Full text of The Southwest Economy : Fourth Quarter 2017

View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

DALLASFED
FOURTH QUARTER 2017

Southwest
Economy

}

Texas Sees Coverage Gains
Under Health Care Act
PLUS

}}
On the Record: Harris County Faces Challenges
Following Hurricane Harvey Deluge

}}
Leading Indicators, Storm Data Guide Houston
Economic Forecast

}}
Mexico’s ‘SOFOM’ Finance Firms Attempt to Broaden
Loan Availability

}}
Spotlight: Rising Education Helps Explain Hispanic
Household Income Growth in Texas

PRESIDENT’S PERSPECTIVE

headline
}The
unemployment rate
and other measures of
labor market utilization
are at or below
prerecession lows.

s a member of the Federal Open Market Committee, I am particularly focused on monetary policy
and achievement of the Fed’s dual-mandate objectives of full employment and price stability. In
the pursuit of these objectives, I voted in March and again
in June of this year to increase the federal funds rate. I also
supported the September decision to begin the process of
gradually reducing the size of the Federal Reserve balance
sheet.
The decision to remove monetary policy accommodation is, at least in part, based on my judgment that the
U.S. economy is at or near full employment. The headline
unemployment rate and other measures of labor market
utilization are at or below prerecession lows. In addition,
the Texas unemployment rate has now slipped below 4
percent, a level we have not experienced in several decades.
Cleanup and rebuilding after Hurricane Harvey is adding to job growth and tightening labor markets in Texas. As
Jesse Thompson writes in this issue of Southwest Economy,
the storm should not materially reduce the rate of Houston
job growth, which is expected to reach 2 percent in 2018.
The economy can grow through increased employment and also by making the workforce more productive.
While education is a key driver of worker productivity,
a healthier workforce can also play an important role in
increasing productivity. In this issue’s cover article, Anil
Kumar discusses Texas’ experience with the Affordable
Care Act. Although Texas opted out of the Medicaid expansion, the state’s insured rate still jumped 6 percentage
points from 2013 to 2016—a development that, if sustained,
should have positive consequences for the productivity of
the Texas workforce.
Federal Reserve Bank of Dallas economists are actively
doing research that gives us greater insight into economic
conditions in this district and the nation. We actively disseminate this research in publications such as Southwest
Economy in order to inform policymakers and the public.
The work of our economists suggests that, even though
Harvey dealt a severe blow to the state, we will rebound
from this storm and should resume strong growth in the
years ahead. I am very optimistic about the future prospects
for the region in 2018.

Robert S. Kaplan
President and Chief Executive Officer
Federal Reserve Bank of Dallas
December 11, 2017

Texas Sees Coverage Gains
Under Health Care Act
By Anil Kumar

}
ABSTRACT: While Texas was
among the states choosing
not to participate in the
Medicaid expansion under
the Affordable Care Act,
it nonetheless has seen
improvement in the share of
the population with health
insurance coverage. Gains
are notable among the noncollege-educated workingage population in Texas, a
state that has long ranked
near the bottom in health
care coverage nationally.

exas, with one of the nation’s
most vibrant economies, has
historically ranked among
the states with the highest
uninsured populations.
The gap between Texas and other
states had narrowed steadily until the
Affordable Care Act (ACA) took effect
in 2014. After the state decided to opt
out of the ACA’s far-reaching Medicaid expansion, the gap again widened
(Chart 1).
A closer look at the data before
and after ACA implementation reveals
that the uninsured rate declined significantly in Texas due to an increase
in private health insurance coverage.
Nationally, however, the rate reduction
was larger.
The Texas uninsured rate remains
elevated among several key demographic groups, and increases in coverage could have been larger had the
state opted to expand Medicaid.
Assessing the ACA’s relative impact in Texas provides useful insights
into the insurance market, even amid

Chart

continuing attempts in Washington to
repeal the health care law and roll back
the Medicaid expansion.

Qualifying for Medicaid
Medicaid is a means-tested public
health insurance program for lowincome individuals—mainly families
with children, pregnant women, the
elderly and the disabled. The program
is jointly funded by federal and state
governments but administered by the
states under federal rules.1 It is the
largest means-tested transfer program
in the U.S. and has experienced rapid
long-term expenditure and enrollment
growth.
Medicaid expenditures account
for about 10 percent of federal spending, up from 2.4 percent in 1980.2
Following the program’s inception in
1965, eligibility was traditionally tied
to receipt of welfare assistance. The
program covered mainly single women
with children on cash assistance, and
low-income elderly people receiving
Supplemental Social Security Income.

Uninsured Rate Declines Under ACA; U.S.–Texas Gap Widens

Percent

U.S.
26.2

26.3

25.7

25.4

25.0

Texas

24.6
21.3

20
16.5

17.2

17.7

19.1
17.3

16.9

18.6

16.7

13.5
10.9

2008

2009

2010

2011

2012

2013

2014

2015

10.0

2016

NOTES: Data reflect the percentage of the civilian noninstitutionalized population under 65 that is uninsured. ACA refers
to the Affordable Care Act.
SOURCE: Census Bureau, American Community Survey one-year estimates.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

A series of expansions in the late
1980s and the 1990s extended Medicaid to other low-income individuals who did not meet more stringent
requirements for traditional cash
assistance—pregnant women, and
parents with children. But Medicaid
eligibility remained strongly linked to
family structure, with the program in
most states out of reach for nondisabled, nonelderly adults without minor
children, regardless of income.
Medicaid differs from Medicare—
the health insurance program, financed
by federal payroll taxes, for all senior
(65 and older) and disabled people who
are eligible for Social Security benefits.
Medicare beneficiaries with low income
are additionally eligible for Medicaid for some health care services not
covered under Medicare—for example,
long-term nursing home care beyond
the 100 days covered by Medicare.3

Changes Under ACA
In the most sweeping Medicaid
expansion since the program’s inception, the ACA as signed into law in 2010
required states to extend Medicaid eligibility to all nonelderly adults—regardless of disability or family structure—
whose incomes were up to 138 percent
of the federal poverty line (FPL). (In
2017, the FPL for determining Medicaid
eligibility was $20,420 for a family of
three, increasing by about $4,200 for
each additional family member.)
However, after a 2012 Supreme
Court ruling made additional Medicaid
coverage optional for states, only 32
states (and the District of Columbia)
opted in.4 Texas was one of the 18 states
to opt out and, thereby, forego more
generous federal matching of state
costs to cover additional beneficiaries
under the ACA expansion. The expansion called for a 100 percent match
from 2014 to 2016, gradually declining
to 90 percent in 2020 and beyond.5
The ACA also dramatically overhauled the private insurance market.
The law instituted an “individual mandate” requiring that most Americans
have health care coverage (or face a tax
penalty). It also established an “employer mandate” stipulating that employers

with 50 or more full-time-equivalent
workers offer affordable health insurance to employees (or pay a fee).
The “dependent-care mandate,”
a provision that took effect in 2010,
compelled health insurance companies
to allow parents to obtain coverage for
dependents up to age 26. Another provision enabled workers without access
to qualified employer-provided health
care coverage to purchase insurance
through an ACA-sponsored marketplace. Consumers with incomes of 100
percent to 400 percent of the FPL are eligible for a tax credit on health insurance
plan premiums (premium subsidy), and
those with incomes of 100 percent to
250 percent of the FPL are additionally
eligible for assistance with out-of-pocket costs (cost-sharing subsidy).

Lesser Benefits in Texas
Even before the ACA’s arrival,
Texas tightly limited Medicaid eligibility for most demographic groups. While
income thresholds for children and
pregnant women to qualify are close
to the national average, the eligibility
standards for nonelderly parents have
lagged significantly behind the rest of
the nation.
In 2013, before the ACA took full
effect, a nonelderly parent with a family of three in Texas needed a family
income less than 25 percent of the FPL
to qualify for Medicaid. The national
average was 87 percent.6
With the ACA’s Medicaid provisions, the eligibility cutoff rose to 138
percent of the FPL. But the cutoff fell to
just 18 percent of the FPL in Texas after
the state opted out of the expansion.
The national average rose to almost 100
percent of the FPL.7
Texas’ eligibility qualifications
for children and pregnant women
are much more generous relative to
those for parents and are closer to the
national average.
Like other states, Texas is required
to extend Medicaid coverage to lowincome elderly people who also are
eligible for the Supplemental Social
Security Income program, which has
an income eligibility limit of 74 percent
of the FPL. Unlike 33 other states, Texas

does not have a medically needy program for elderly people with incomes
higher than the Medicaid eligibility
limit. The medically needy program allows seniors with high medical expenses, but with income above Medicaid
eligibility limits, to qualify for Medicaid
by spending down their household
resources on medical expenses.

Medicaid Changes Under ACA
The Medicaid coverage rate for the
nonelderly population in Texas was
relatively high prior to the ACA—17.6
percent in Texas versus 18.6 percent for
the nation.
In addition to differences in demographics and income distribution,
higher Medicaid coverage among Texas’
children kept the gap with the U.S.
small, despite Texas’ near-bottom ranking among states in Medicaid generosity
for key demographic groups. The share
of children on Medicaid was 39 percent
in Texas versus 37 percent for the U.S.
Enrollment in Medicaid and the
Children’s Health Insurance Program
(CHIP) rose 38 percent in Medicaid-expanding states between July–September 2013 and July 2017. Nonexpanding
states also experienced a 12 percent
enrollment increase, partly due to the
ACA raising awareness of the program
among Medicaid-eligible households
that hadn’t previously participated.
Enrollment rose 6.9 percent in Texas,
compared with 29 percent nationally.8
Not surprisingly, a significant U.S.–
Texas gap in Medicaid coverage of the
nonelderly population emerged after
the ACA. While coverage remained
largely flat in Texas at about 18 percent
of the nonelderly population, it rose 3
percentage points nationally. Roughly
22 percent of all nonelderly Americans had received health care coverage through Medicaid as of late 2016
(Chart 2).

More Private Coverage in Texas
As the U.S.–Texas gap in Medicaid coverage widened, the state and
national gap narrowed for those with
insurance, largely due to the ACA’s
overhaul of the private insurance market that applied to all states (Chart 3).

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Individuals with employer-based
coverage increased 3 percentage points
from 2013 to 2016 in Texas—from 51 to
54 percent. By comparison, that share
nationally rose about 1 percentage
point, to 59 percent.
Employer-based insurance remains
the mainstay of the U.S. health insurance
system because the workplace provides
an efficient mechanism to pool health
insurance risk. If health insurance is optional, individuals with high health risks
are more likely to purchase coverage.
When insurers are unable to
determine the exact health status of
individual prospective policyholders,
they tend to charge high premiums for
directly purchased insurance or may
not cover preexisting conditions—an
attempt to minimize potential losses.
Thus, the cost of private, nongroup insurance is substantially higher than for
employer-based group plans. Through
the individual mandate and the health
insurance marketplace, the ACA attempted to create a diversified risk pool
for nongroup private insurance.
About 1.2 million Texans were
enrolled in an ACA marketplace health
insurance plan during 2017, with
83 percent eligible for premium tax
credits and 61 percent qualifying for
cost-sharing subsidies.9 Insurance from
all private sources (including employers) increased 7 percentage points in
Texas—compared with a 4-percentagepoint gain nationally.
Increases in both Medicaid and
private insurance coverage at the
national level suggest that the Medicaid expansion didn’t simply crowd
out private insurance. A substantial
crowd-out can neutralize much of the
gain from increased Medicaid coverage
if beneficiaries drop private coverage in
favor of Medicaid.
Significant declines in the uninsured rate among the nonelderly suggest that the crowd-out was small. The
uninsured rate fell 7 percentage points
to 10 percent nationally and 6 percentage points to 19 percent in Texas.
The nonelderly population includes
children and people below age 26 who
benefited from the dependent care
mandate of the ACA.

Chart

Medicaid Coverage Jumps in U.S. as ACA Takes Effect

Percent

24
22.1

22
20
U.S.
18

17.9

Texas

16
14
12

2008

2009

2010

2011

2012

2013

2014

2015

2016

NOTES: Data reflect the percentage of the civilian noninstitutionalized population under 65 with Medicaid. ACA refers to
the Affordable Care Act.
SOURCE: Census Bureau, American Community Survey one-year estimates.

Chart

Private Health Insurance Coverage Climbs Under ACA

Percent

69.1
U.S.

63.9

Texas

2008

2009

2010

2011

2012

2013

2014

2015

2016

NOTES: Data reflect the percentage of the civilian noninstitutionalized population under 65 with private health insurance. ACA refers to the Affordable Care Act.
SOURCE: Census Bureau, American Community Survey one-year estimates.

Non-College-Educated Groups
Focusing on the nonelderly
population over age 26 with no college
education can provide more precise
estimates of the effects of broadening
Medicaid eligibility. The non-collegeeducated population would have
been more intensely affected by the
Medicaid expansions. Previously, the
uninsured rate among this group was
43 percent in Texas and 29 percent
elsewhere in the U.S.10
Comparing the states that expanded Medicaid and those that did
not also helps in the analysis. Medicaid
coverage among those without college

increased 8 percentage points in states
that expanded Medicaid but just 2
percentage points in states that did not
expand (Chart 4A). Assuming other
factors followed similar trends, the difference of 6 percentage points can be
largely attributed to the expansion.
While employer-provided coverage remained virtually unchanged in
both groups of states, private coverage
increased almost 5 percentage points
in expanding states and 6 percentage
points in nonexpanding states.
Thanks to negligible crowd-out
from Medicaid expansion, the uninsured rate for the non-college group

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Chart

Affordable Care Act’s Expansion Boosts Coverage for Non-College-Educated Individuals

A. Medicaid Coverage Among Non-College Educated Increases
More in Medicaid-Expanding States

B. Uninsured Rate Among Non-College Educated Drops More
in Medicaid-Expanding States

Percent

2011–13

26.7

2014–16

25
20
15

14.5
8.7

10.8

43.3

2014–16

35.5

34.2
27.2

25
20

28
17.1

15
10
5

5
0

2011–13

45
40
35

18.6
12.5

Nonexpansion states

Expansion states

Texas

Nonexpansion states

Expansion states

Texas

NOTES: The sample was restricted to 27–64-year-olds with no college education. Expansion states include those that expanded Medicaid coverage effective Jan. 1, 2014.
SOURCES: Census Bureau, CPS-IPUMS, March supplement; author’s calculations.

dropped 11 percentage points in
expanding states, compared with 7 percentage points in nonexpanding states
(Chart 4B). Thus, the additional decline
of 4 percentage points in the uninsured
rate in the expanding states could potentially be tied to the expansion.
Texas, without broader Medicaid
coverage, benefited from changes in the
private insurance markets through the
ACA. While Medicaid among non-college-educated adults increased about
2 percentage points, private insurance
coverage jumped 7 percentage points.
The uninsured rate for this group fell 7
percentage points to 36 percent.
Despite improvements, the uninsured rate remains elevated across
key demographic groups in Texas and
elsewhere in the nation (Table 1). The
gap is particularly wide among the
non-college educated. Lower Medicaid
coverage across the board in Texas is a
primary reason.

Law’s Economic Impact
Medicaid expansions and the
ACA’s subsidies, which led to increased
health care coverage, came at a cost to
taxpayers. The Congressional Budget
Office (CBO) projected a net price
tag of $1.4 trillion between 2017 and
2026.11 An important component of
that is the negative impact on work effort, namely employment and hours.

Researchers have understood that
expanding entitlement programs such
as Medicaid can have important implications for the labor market. The most
basic effect on such outcomes—employment, work hours and earnings—is
similar to increasing wealth or income.
If eligible low-income individuals
value Medicaid and think of it as more
income, they tend to work less, just like
anyone else who feels wealthier.
Besides income effects, the
income eligibility cutoffs create other
incentives for changing the employment and work hours of those who are
close to benefit thresholds. Those just
above the limit might reduce earnings
to qualify for Medicaid; those below
the new limit would be open to work
more and increase earnings because
they can still qualify for Medicaid.
Availability of ACA marketplace
subsidies for nonelderly adults starting
at 100 percent of the FPL and gradually
phasing out at 400 percent of the FPL
widens the scope of workers that might
adjust their incomes to maintain eligibility for those subsidies. The reduction
in subsidies with higher earnings acts
as an effective tax on additional work.
Also, the ACA’s employer mandate may
induce some employers to rely more
on part-time workers.
Moreover, many low-income individuals may hold regular full-time jobs

simply to maintain employer-based
health insurance. Medicaid eligibility
may prompt these people to give up
full-time jobs and opt for lighter and
more flexible schedules with fewer
hours. Some could retire early if Medicaid were available before age 65.
Such behavioral effects suggest
that Medicaid expansion should lower
labor force participation, employment
and hours worked. The CBO estimates
that various provisions of the ACA
would lower total hours worked 1.7
percent and total earnings about 1
percent by 2025; there would be 2 million fewer full-time-equivalent workers
in 2025 than would be the case without
the ACA.12
At a time when labor force growth
is already projected to slow due to an
aging population and retiring baby
boomers, ACA-related employment
declines could be a further drag on
growth. Nevertheless, some positive
spillovers from increased health care
coverage helped limit the CBO’s estimate of reduced employment.
First, some individuals stay with
their employers simply to maintain
insurance even though they could
be more productive elsewhere, and
quitting could render them uninsured
until they find another job. Availability
of public insurance coverage through
Medicaid should reduce such an

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Table

panded coverage. Because low-income
individuals spend a relatively larger
share of additional income than higherincome households do, the net effect of
the redistribution on consumer spending could be modestly positive.17

Health Insurance Coverage Lags in Texas,
Especially Among the Non-College Educated

A. Percent with Medicaid Coverage Before and After the ACA, by Race and College
White

Black

Hispanic

Pre-ACA Post-ACA

All
Pre-ACA

Post-ACA

College
U.S. (ex. Texas)

4.9

7.8

11.2

14.2

10.3

15.4

6.2

9.4

Texas

2.5

3.8

5.3

7.9

3.7

4.6

3.1

4.4

13.2

17.8

24.6

29.1

18.8

26.6

16.5

22.1

7.0

8.9

17.4

21.0

8.3

9.9

8.7

10.8

Non-college
U.S. (ex. Texas)
Texas

B. Percent Uninsured Before and After the ACA, by Race and College
White

Black

Hispanic

Pre-ACA Post-ACA

All
Pre-ACA

Post-ACA

College
U.S. (ex. Texas)

12.8

9.5

20.5

15.1

24.5

14.9

14.8

10.7

Texas

16.2

12.3

23.4

17.6

28.6

21.9

20.4

15.1

U.S. (ex. Texas)

22.1

15.6

30.2

19.6

44.3

29.2

28.7

19.6

Texas

28.2

24.2

34.3

27.2

53.7

44.2

43.3

35.5

Non-college

NOTES: Sample restricted to those 27–64 years old. ACA refers to the Affordable Care Act.
SOURCES: Census Bureau, CPS-IPUMS, March Supplement; author’s calculations.

“employment lock” and make the labor
market more efficient.
Second, Medicaid expansion
through increased income eligibility
limits could lead to reduced welfare
caseloads among individuals who
maintained welfare eligibility simply to
qualify for Medicaid. With enhanced
limits, they may be drawn into the
labor market because they could still
qualify for Medicaid. Previous research
has found compelling evidence of the
positive effects of Medicaid expansions
on the “welfare lock.”

Employment, Consumer Spending
Employment data before and during the ACA that compares Medicaidexpanding and nonexpanding states
suggests the employment rate was little
changed even for the most-affected
individuals—non-college-educated
adults—in the two groups of states. Other detailed research has reached similar
conclusions.13 Except for select groups,
such as childless adults and dependents
who benefited from the dependent care
mandate, the Medicaid expansions
have largely been neutral with respect
to key labor market outcomes.

Other ripple effects of more widely
held insurance also help offset the cost
to taxpayers. Lack of health insurance
is a key driver of financial distress for
those without coverage. Not surprisingly, increases in Medicaid coverage
are strongly associated with lower personal bankruptcy rates.14 The Medicaid
expansions and ACA’s marketplace
subsidies should ease financial stress
among low-income people who obtain
health care coverage.
Without such coverage, the uninsured can’t pay for their hospital stays
and emergency room visits, shifting
the cost to the insured through higher
insurance premiums and to taxpayers
through higher levies. Such uncompensated care costs have declined
following ACA implementation.15
Expanded health care coverage
also boosts consumer spending by
limiting the need for precautionary
saving to meet the out-of-pocket costs
of unforeseen medical expenses among
potentially eligible households.16
Increased spending among those with
health coverage could be partly offset
by reduced consumption among those
facing higher taxes to fund the ex-

Remaining Challenges
Although Texas opted out of the
Medicaid expansion, the uninsured
rate in the state fell among major demographic groups because of sharply
higher private insurance coverage.
Challenges remain, however, as the uninsured rate for some groups remains
elevated and the gap between Texas
and the nation has increased.
Thus far, there appears little
evidence of negative effects on the
labor market in states that participated
in Medicaid expansion. Whether the
large gains in health coverage are
worth the budgetary cost remains an
open question.

Kumar is an economic policy advisor
and senior economist in the Research
Department at the Federal Reserve
Bank of Dallas.
Notes
1
The federal share of state Medicaid costs is governed
by the federal matching assistance percentage—a
formula based on a state’s per-capita personal income
relative to the nation—and ranges from 50 percent to 74
percent, with lower per-capita income states receiving a
higher share.
2
See “Trends in Medicaid Spending,” Medicaid and
CHIP Payment and Access Commission, June 2016,
www.macpac.gov/wp-content/uploads/2016/06/Trendsin-Medicaid-Spending.pdf.
3
See Centers for Medicare and Medicaid Services for
more details on the Medicare-Medicaid relationship,
www.cms.gov/Research-Statistics-Data-and-Systems/
Statistics-Trends-and-Reports/CMSProgramStatistics/
index.html.
4
See “Status of State Action on the Medicaid Expansion
Decision,” Kaiser Family Foundation, Nov. 8, 2017, www.
kff.org/health-reform/state-indicator/state-activity-aroundexpanding-medicaid-under-the-affordable-care-act.
5
For further details on the fiscal impact of the decision
on Texas, see “Texas Health Coverage Lags as Medicaid
Expands in U.S.,” by Jason Saving and Sarah Greer,
Federal Reserve Bank of Dallas Southwest Economy,
Fourth Quarter, 2015.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

(Continued on back page)

ON THE RECORD
A Conversation with Judge Ed Emmett

Harris County Faces
Challenges Following
Hurricane Harvey Deluge
Edward M. Emmett became Harris County judge in 2007. He
is the chief administrative officer and director of emergency
management in the county, which includes most of the city of
Houston. He recently released a 15-point plan to prevent future
flooding disasters. Harris County is the third-most populous U.S.
county, accounting for two-thirds of the Houston metropolitan
statistical area’s population of 6.8 million people.

Q. How did Hurricane Harvey’s aftermath compare to previous severe
weather events in the region?
There is no comparison—Harvey
is by far the worst storm to hit Harris
County. Unlike past events such as Hurricane Ike, Harvey was a rain and flooding event that affected a much greater
number of people and businesses.
Over 50 inches of rain fell in parts
of the county; there is very little you can
do to prepare for that amount of rain in
a short period of time. With a hurricane,
there is a storm surge that is localized
and more predictable, which allows you
to better prepare and evacuate people.
What many people don’t understand is
that Harris County has good drainage—
that’s why most of the water was gone
within a week. It was just too much rain
in a very short period of time, and for
homeowners, it has been a much more
difficult event to deal with than businesses since homeowners don’t have the
resources to rebuild. Going forward, the
biggest challenge is finding the money
to rebuild and beef up infrastructure to
reduce the impact of the next big flood.

Q. What do you see as key differences in the response to Harvey
compared with Superstorm Sandy
and Hurricane Katrina?
Katrina was a game changer in
terms of how large a political event it
was and how governmental entities
reacted in the aftermath. There was a
lot of criticism of federal, state and local governments. What a lot of people
remember is the disjointed response.
By the time I had become county
judge in 2007, the precedent had already
been set in our region that during these
kinds of crises, the city, county and state
need to work together. We don’t get
caught up in who is in charge of what,
we simply do what needs to be done to
make sure everyone is safe.
Hurricanes didn’t used to be political events. As an example, I was in the
Houston area when Hurricane Alicia hit
us in 1983, and it wasn’t an event that
came into the political realm. Nobody
talked about the government’s response
or how FEMA (the Federal Emergency
Management Agency) managed the
aftermath. That first changed with the
politicization of Hurricane Andrew [in
South Florida] in 1992—an election

year—when President [George H.W.]
Bush was judged by how he responded.
Since then, these storms have become
political in the sense that the response
to the event is judged and used as ammunition in the next election cycle.

Q. Businesses have told us that
Harvey did not cause as much business disruption as it did residential
disruption. How does this impact
the recovery?
Businesses have the resources
to start the repairs right away, and
most were back on their feet relatively
quickly. Even a small restaurant I know
of in Meyerland, one of the hardest-hit
areas in the county, took on five feet of
water but was back in operation within
three weeks. Once the water was gone,
businesses had the wherewithal to begin
rebuilding and get back into operation.
The issue with homeowners is
that most people have a significant
money shortage and don’t have the
funds to rebuild. Many homeowners
were not insured, and even those who
were are waiting a long time for FEMA
to send them checks. Even then, often
the amount received doesn’t cover the
cost to rebuild. So, many have been left
waiting for additional aid or hoping for
a buyout.

Q. What do you see as the most
important points of your recently
announced flood control proposal?
What are the biggest challenges?
The most important element of the
plan is the overall vision. We need to
acknowledge that we live in a floodprone area and take action to reduce
the impacts of future floods. Rather than
fighting with our watersheds, we need to
use them as assets and turn as many of
them as possible into recreational areas
and green spaces. It is a different kind of
mindset that we need to adopt.
Beyond that, we need to recognize
that lakes Houston and Conroe need
to be designated as flood-control lakes
rather than water supplies. We need to
change our thinking and think of everything as a flood-control effort.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

}prone areas, where they really shouldn’t have

SPOTLIGHT
If people have been allowed to build in floodbuilt, we need to buy them out so that we don’t
keep paying out insurance.

The biggest challenge is finding
the money. The federal dollars will
flow sooner or later, but they won’t be
enough. Everyone agrees we need a third
reservoir on the west side, but the estimated cost is $500 million. With just 5
percent of the rainy-day fund [the Texas
Economic Stabilization Fund] balance,
the state could cover the cost and protect
a huge number of Texas residents without waiting for federal funding.
An important use of funds would
be to buy out homes in true flood plains.
If people have been allowed to build
in flood-prone areas, where they really
shouldn’t have built, we need to buy
them out so that we don’t keep paying
out insurance.

Q. How does the governmental structure of the region impact its infrastructure planning to prevent damage during future severe weather?
We need a long-term revenue
source that encompasses unincorporated Harris County to finance these
infrastructure projects.
A huge number of people live
outside of incorporated areas of Harris
County. Compared with Dallas County,
where there are about 6,000 people in
unincorporated areas, there are almost
2 million in Harris County—nearly the
same as the city of Houston’s population. We have more than 1,000 different municipal utility districts in Harris
County. Because of (state) legislation,
the city of Houston can’t annex these areas. So, we have an area with a growing
urban population that expects the same
services as the city but limited avenues
for obtaining revenue to provide them.
At the county level, we only have
access to property taxes and not sales
taxes. There is a lot of pressure at the

state level to not only keep property
taxes from increasing but to reduce
them. The county government is an arm
of the state—we can only do what the
state tells us to do. But this pressure to
maintain services, including flood control to a large and growing population,
while at the same time facing cuts to our
revenues makes it a difficult balance.
A state Senate bill considered in the
last legislative session proposed restricting county revenue to the pace of population growth and inflation. While this
might sound good, this doesn’t realistically work for a place like Harris County
where most of the growth is in unincorporated areas. On top of that, the county
government’s responsibilities include
indigent health care, criminal justice,
roads and bridges, and flood control,
which aren’t well-tied to any measures
of inflation that I know of.
Ultimately, we need some way to
find a more sustainable source of revenue, such as sales taxes, to help fund
some of these projects.

Q. What are the main issues impacting the region’s future?
Transportation is key going forward. The Houston region is the gateway
of North America for international trade.
We need to find a way to move freight
more efficiently throughout the state.
The highways in and around Harris
County are getting very congested, and
improvements need to be made if we
are to capitalize on our advantages in
this arena.
In terms of Harris County, mass
transit will come since simply adding
more and more highways is not a viable
long-term solution. However, the way
the area has developed over time means
that traditional fixed commuter rail isn’t

a very practical solution. Harris County
is big but not nearly as dense as many of
the other large cities in the U.S. This region is very large and as population has
grown, people moved into the suburbs
for schools and affordable housing. In
general, rail is not flexible to move with
the demand as workers and companies
relocate and evolve in the region.
That said, I think whatever solution
ultimately evolves will include some
commuter rail, and there have been
opportunities missed in the past. In particular, the old Missouri–Kansas–Texas
Railroad included a rail line coming
from Katy directly to downtown, which
would have been a great piece to include in a broader mass transit solution.

Q. How do you see the aftermath of
Harvey affecting the future regional
economy? Has it been traumatic
enough to hamper medium- to longterm growth?
I think it is too early to tell. There
have been three 500-year floods in the
past two years. Obviously, we have a
problem with what our definition of
a 500-year flood is, because we can’t
assume we’ll go another 1,500 years
without a significant flood. We need to
start over and redefine our flood plains.
The bigger issue facing the economy here in terms of future growth is the
national and global perception of Houston and Harris County. How do people
outside of Houston perceive the area as
a place to live? While only about 5 to 7
percent of Harris County homes were
damaged, there is a narrative out there
that the area was totally inundated and
that homes and businesses are likely
to flood. A lot of our conversations are
about how to counteract that narrative
and push the positives of the region.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Leading Indicators, Storm Data
Guide Houston Economic Forecast
By Jesse Thompson

}
ABSTRACT: A forecasting
model for Houston that
incorporates storm damage
data and leading economic
indicators can help project
post-Hurricane Harvey
employment growth. The
forecast indicates that
Houston’s economy will grow
near its 2 percent historical
average in 2018.

surging energy sector helped
Houston metropolitan
employment expand at an annual rate of nearly 4 percent
from 2011 through 2014—the equivalent of 325,000 jobs over the period.
But in the subsequent two years, job
growth stalled as rising crude production drove down oil prices.
A recovering energy industry
helped propel Houston to above-trend
growth in the first half of 2017 before
Hurricane Harvey walloped the region
in late August. The destructive storm
disrupted economic activity, bringing with it a challenge for economic
forecasters.
Businesses rely on job growth
projections to plan for capital expenditures as well as more basic requirements such as office space, staffing
and vehicle demand. Four economic
models have been developed that rely
on past job growth and leading indicators to forecast Houston employment
growth, including a new experimental
leading index.
Incorporating the dollar cost of
direct storm damage improves model
accuracy, and averaging the independent model forecasts tends to produce
more accurate longer-term predictions.
These models taken together anticipate
that after three months of rapid recovery from Hurricane Harvey, Houston
will grow near its historical average rate
of 2 percent in 2018.

High Growth, Volatility
The Houston metro area was the
second-fastest growing of the nation’s
20 largest metros from 1990 to 2016,
adding jobs at an annual rate of 2.2
percent, just behind Dallas–Fort Worth
at 2.3 percent. That compared with a
national average growth rate of 1 percent during the period.

Houston had more than 3 million jobs in 2016, accounting for 2
percent of all U.S. payroll jobs, and a
gross domestic product of $478.6 billion, amounting to 2.6 percent of U.S.
output. Houston is home to about onefourth of all Texas jobs, nearly a third of
the state’s output and almost a quarter
of the state’s population.
Despite the metro area’s heft,
Houston’s high growth comes with
volatility. The area experienced the
most volatile job growth of the eight
largest U.S. metros from 1990 to 2016.
It was also the nation’s fifth-mostpopulous metro area in 2016, with 6.8
million inhabitants (Table 1).
Oil prices are responsible for much
of Houston’s volatility. They affect oil
producers’ revenues and future drilling
activity. The supply chains for most U.S.
oil and gas operations have connections
to Houston—an industry headquarters
city—although little oil is being produced in the immediate area. Houston
retains the title “energy capital of the
world,” despite diversification and
deepening connections to the broader
U.S. economy over the past 30 years.1
Additionally, many local businesses and residents also own mineral
rights and receive royalty payments
from oil and gas production.2 Industries
not generally associated with the energy
sector, such as business and professional services, have direct and indirect
connections to energy. Thus, Houston’s
service sector employment (excluding
government) is the second-most volatile among the largest U.S. metros.
Forecasts by their very nature assume that the behavior of data in the
recent past will carry into the future.
How can businesses plan ahead given
recent volatility? Tools are needed that
help capture the sources of that volatility and identify underlying trends.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Data and Leading Indicators
The Bureau of Labor Statistics
(BLS) produces the most commonly
used measure of regional employment
growth. The Current Employment
Statistics (CES) jobs data are generated
from a national survey of 634,000 worksites, covering about one-third of total
nonfarm civilian jobs. Smaller sample
sizes at the metro level can result in
significant changes when the BLS annually revises its estimates based on
unemployment insurance data, which
are comprehensive but have lagged
availability.
The Federal Reserve Bank of Dallas works to improve the accuracy of
recent employment data in Texas and
its major metros through a quarterly
early benchmarking process and a
two-step seasonal adjustment.3 These
processes help make the Dallas Fed’s
local employment data more accurate in real time than unrevised CES
estimates while taking into account
seasonal variations (like more retail
workers before Christmas) that can
obscure trends.
Recent job growth numbers set the
trend for most forecast models, so boosting the accuracy of those data should
improve ensuing predictions. Leading
economic indicators when combined
with improved employment data presumably better capture the cyclicality
and volatility of future job growth.
Among the useful indicators that
contain information about Houston’s
near-future job growth and likely
business-cycle changes are the Houston
Purchasing Managers Index (HPMI)
produced by the Institute for Supply
Management (ISM) and the Texas Leading Index (TLI) from the Dallas Fed.
Both are powerful barometers of impending changes in the local economy.
The HPMI is a monthly diffusion index that is similar to the ISM’s
national purchasing managers index.
Supply managers from a broad group
of industries, including health, manufacturing, oil and gas, and services
answer questions seeking to ascertain whether business conditions are
improving, worsening or unchanged
relative to the prior month. Responses

Table

Houston Job Growth Volatile Relative to Other Large Metros
Total
nonfarm
jobs
Rank

United States

Service-providing
nonfarm jobs
(ex. government)

Volatility

Rank

Population
in
2016

Volatility

Millions

1.03

323.1

0.96

Houston-The Woodlands-Sugar Land

1.45

1.43

6.8

Miami-Fort Lauderdale-West Palm Beach

1.43

1.44

6.1

Los Angeles-Long Beach-Anaheim

1.31

1.14

13.3

Dallas-Fort Worth-Arlington

1.27

1.37

7.2

Washington-Arlington-Alexandria

1.05

1.24

6.1

New York-Newark-Jersey City

1.03

0.94

20.2

Chicago-Naperville-Elgin

1.02

0.92

9.5

Philadelphia-Camden-Wilmington

0.93

0.89

6.1

NOTES: Volatility is calculated as the standard deviation of the absolute 12-month log-change in employment from January
1991 through December 2016. A larger standard deviation means the 12-month growth rate is more variable.
SOURCES: Bureau of Labor Statistics; Census Bureau.

are compiled into eight component
indexes—sales, production, employment, purchases, prices paid, lead
times (from sellers), purchased inventory and finished goods in inventory.
The responses are then combined into
an index in which a value above 50
indicates an expanding economy and a
value below 50 suggests contraction.
Since its inception in early 1995, the
index has consistently provided an early

Chart

indication of changes in employment
growth rates and turning points in the
broader regional economy. It is also very
timely, typically available on the 10th
day following the measured month.
Whenever the HPMI strengthens or
weakens, job growth over the next few
months most likely follows (Chart 1).
The Dallas Fed’s TLI is a different
kind of leading index. It combines eight
separate indicators associated with

Job Growth Tracks the Houston Purchasing Managers Index

Percent

Index value

70
Houston job growth

50
HPMI

–2

–4

–6

–8

30
1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

NOTES: Employment growth is the three-month percent change in the centered moving average. The
Houston Purchasing Managers Index (HPMI) is depicted as a centered three-month moving average where values >50
indicate expansion.
SOURCES: Institute for Supply Management; Bureau of Labor Statistics; adjustments by the Federal Reserve Bank of Dallas.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

future business activity that typically
change direction three to nine months
before the rest of the economy does.
For example, rising initial claims
for unemployment insurance suggest
firms believe they may be unable to
support staffing levels; individuals
losing their jobs will likely scale back
consumption in the months ahead.
When help-wanted advertisements
rise, employers are more confident
about their outlook and plan to hire
more staff. As those positions are filled,
new employees are likely to increase
consumption in future months.
Both indicators are included in the
TLI. Other items are the Texas value of
the dollar, a trade-weighted index that
accounts for inflation; the U.S. leading index from the Conference Board;
the real (inflation-adjusted) price of
West Texas Intermediate crude oil; oil
and natural gas well permits; a Texas
stock index representing the 100 largest
publicly traded companies based in
the state, and average weekly hours
worked in manufacturing.4
The TLI is the main component
of the Dallas Fed’s Texas forecasting
model, which has consistently outperformed other state-level employment
forecasts tracked by the Western Blue
Chip Economic Forecast.5

Chart

An index constructed to help forecast the Texas economy should have
predictive power for Houston. Analysis
suggests that the TLI is significantly correlated with Houston job growth one to
six months into the future (Chart 2).

Houston Leading Index
The TLI and the sales and production components of the Houston
Purchasing Managers Index are subsequently combined with data covering additional metrics to produce an
experimental index of leading indicators for Houston (HLI).
The additional data are: HelpWanted OnLine advertising, singlefamily housing construction permits,
existing-home sales, the American
Chemistry Council’s U.S. chemical production index, the Bloomberg Houston
150 stock market index, the average
monthly price of West Texas intermediate crude oil, the U.S. rig count and the
Conference Board’s U.S. index of leading economic indicators.6,7
The new index’s construction
resembles the TLI and the U.S. leading index. Changes in each of the 11
components are divided by a measure of their own volatility to prevent
the effects of inherently more noisy
components—such as oil prices—from

Texas Leading Index Correlated with Houston Job Growth
Up to Six Months into Future

Percent

Percent change

35
Houston job growth

6
4

25
15

0
–5

–2

TLI

–15

–4

–25

–6
–8

–35
1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

NOTE: Employment growth and the Texas Leading Index (TLI) are the three-month percent change in a centered moving
average.
SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas.

overwhelming the effects of the others.
These adjusted changes are then averaged to produce a Houston index of
leading indicators.8 Much like the TLI,
the HLI is significantly correlated with
Houston employment growth one to
six months out (Chart 3).

Improving Accuracy
Four different employment forecasting models—three of them based
on measures of ongoing activity in
Texas and Houston—were developed
for Houston. The HLI, the HPMI and
the TLI were each incorporated in
models using two simultaneous equations, where the first equation forecasts
employment growth based on past
changes in employment and the leading indexes, and the second equation
forecasts growth in the leading index
based largely on lagged values of itself.9
A fourth model produced a forecast by averaging the predictions of
many ARIMA (autoregressive integrated moving average) forecasts. These
ARIMA forecasts use only combinations of past job growth to predict future job growth.10 The HLI-based model
tended to be more accurate at charting
the course of employment growth over
the year ahead.11 It did particularly well
at forecasting four to 11 months out.12
Frequently, averaging predictions
from different models can provide
better forecasts than the individual
models. An average of all four of the
models tested tended to be the most
accurate when forecasting 12 months
ahead, reducing forecast error—the
extent to which predicted job growth
differed from actual job growth—by
15.6 percent relative to the ARIMA
model (Table 2).
Including estimates of the direct
cost of damage from major storms over
the past 26 years in the forecast models
improved the accuracy of the forecast
predictions and, in the most recent
instance, provided estimates of Hurricane Harvey’s employment impact.13

Resurgent Economy Anticipated
The average of the four forecast
models predicted a net drop of 30,000
Houston jobs from August to Septem-

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Chart

Houston Leading Index, Future Job Growth
Significantly Correlated

Percent

Percent change

15
Houston job growth

4
5

2
0

–2

–5

Houston leading index

–4

–10

–6
–8

–15
2005

2007

2009

2011

2013

2015

2017

NOTES: The Houston leading index is experimental. Employment growth and the Houston leading index are depicted as
three-month percent changes in a centered three-month moving average.
SOURCES: Bureau of Labor Statistics; adjustments by the Federal Reserve Bank of Dallas; author’s calculations.

ber, assuming that Harvey caused $70
billion in direct damage. Initial estimates put the number at around 22,000
lost jobs. The impact was also likely to
be short lived, corroborating an earlier
Dallas Fed analysis that suggested the
Texas Gulf Coast would recoup jobs
lost due to the storm as recovery efforts
boosted year-end growth.14
(See “On the Record,” a conversation with Harris County Judge Ed Emmett, page 8.)
In short, a host of leading indicators suggest that Hurricane Harvey,
while devastating to many homeowners and small businesses, likely caused
only one month of net job losses in
Houston. Despite slower growth in the
second half of 2017, the region’s longrun economic momentum is unlikely
to be derailed.

Thompson is a business economist
in the Research Department at the
Federal Reserve Bank of Dallas,
Houston Branch.

Fitzgerald and Jeremy G. Weber, Federal Reserve Bank
of Kansas City, Research Working Paper no. 16-12,
November 2016, www.kansascityfed.org/publications/
research/rwp/articles/2016/asset-ownership-windfallsincome-oil-gas-royalties.
3
See definitions of early benchmarking at www.dallasfed.
org/research/basics/benchmark.cfm and two-step
seasonal adjustment at www.dallasfed.org/research/
basics/twostep.aspx.
4
Texas Employment Forecast, Federal Reserve Bank
of Dallas, Nov. 17, 2017, www.dallasfed.org/research/
forecast.aspx.
5
“Revising the Texas Index of Leading Indicators,” by
Keith Phillips and José Joaquín López, Federal Reserve
Bank of Dallas, Southwest Economy, November/
December, 2007, http://dallasfed.org/assets/documents/
research/swe/2007/swe0706b.pdf.
6
The Houston 150 stock index is produced by

Table

Bloomberg. It is a price-weighted index composed of
major companies based in Houston and significant
employers in the area. The U.S. chemical production
index is produced by the American Chemistry Council
to track chemical production activity in the United States
based on industrial production data from the Federal
Reserve. Help-Wanted OnLine data are produced by
the Conference Board from online job postings for
employment in the Houston metropolitan area.
7
Details of the Conference Board methodology can be
found at www.conference-board.org/data/bci/index.
cfm?id=2161.
8
Component series are also seasonally adjusted where
appropriate. Due to limitations in some of the component
data, the Houston index begins in June 2005.
9
Each system of two equations was estimated using
seemingly unrelated regressions.
10
The ARIMA forecast was a weighted average of
many models automatically selected for each of the 84
iterations over the sample period and weighted based on
goodness-of-fit measures.
11
Each model was used to calculate 84 out-of-sample
forecasts beginning in January 2010 and rolling forward
to December 2016. The overall prediction error was
tabulated for the 12-month forecasts, as well as the
prediction error for each step-ahead.
12
The HLI model was specified as follows:
(Equation No. 1) Dln (emp) = b11 Dln (emp)t-3 + b12 Dln
(emp)t-4 + b13 Dln (emp)t-6 + b14 Dln (HLI)t-1 + b15 Dln
(HLI)t-3 + b16 storms + b17 stormst-1 + b18 stormst-2 + b19
recessions + e.
(Equation No. 2) Dln (HLI) = b21 Dln (emp)t-1 + b22 Dln
(HLI)t-1 + b23 Dln (HLIt-2) + b24 stormst-1 + b25 recessions
+ b26 + e.
13
Tropical Storm Allison in 2001 and the 1994 floods
occurred before the HLI model sample period.
14
See “Short-Term Job Growth Impacts of Hurricane
Harvey on the Gulf Coast and Texas,” presentation by
Keith Phillips and Christopher Slijk, Federal Reserve
Bank of Dallas, San Antonio Branch, Sept. 5, 2017, http://
files.constantcontact.com/668faa28001/d7cdfcae-b8614bb2-9cb7-a1f8e361a878.pdf?ver=1505446495000.

Forecast Averaging Produces Better Long-Term Predictions
Percent improvement
in accuracy over ARIMA

Average

15.6

Houston leading index

14.8

Notes

Houston Purchasing Managers Index

12.7

“Diversified Houston Spared Recession … So Far,”
by Jesse Thompson, Federal Reserve Bank of Dallas,
Southwest Economy, Third Quarter, 2015, www.dallasfed.
org/assets/documents/research/swe/2015/swe1503f.pdf.
2
“Asset Ownership, Windfalls, and Income: Evidence
from Oil and Gas Royalties,” by Jason P. Brown, Timothy

Texas Leading Index

6.7

ARIMA

–

NOTES: Data are the percent reduction in the 12-month-ahead root-mean-squared forecast error relative to the ARIMA
(autoregressive integrated moving average) model.
SOURCE: Author’s calculations.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Mexico’s ‘SOFOM’ Finance Firms
Attempt to Broaden Loan Availability
By Michael Perez and Kelsey Reichow

}

ABSTRACT: The market
presence of Mexican
finance companies
known as SOFOMES has
expanded rapidly since the
global financial crisis. The
firms largely operate as
independent outlets and
provide financing to smalland medium-sized companies
as well as to consumers for
larger purchases. Authorities
see SOFOMES as a way to
expand credit to Mexico’s
informal economy.

exicans increasingly rely on
regulated nonbank finance
companies for their credit
needs. Assets at these firms—
known by the Spanish shorthand as
SOFOM ER—have more than doubled
amid a regulatory effort to formalize
and consolidate the industry or sector
beginning in 2013.1 Meanwhile, the
finance companies’ share of the lending market has grown, driven by new
market entrants (Chart 1).
SOFOM ERs specialize in credit,
lease financing and financial factoring services. They offer auto, personal and department store credit and
commercial financing for small- and
medium-sized enterprises. Some serve
as captive finance companies—institutions providing customer credit for
purchases of parent company products—at stores and dealerships.
Others operate as independent
lenders or as off-balance sheet ve-

Chart

hicles for larger banks. A total of 49
SOFOM ERs operated in Mexico as of
August, with about 614 billion pesos
($33 billion) in loans outstanding.2
Not all Mexican finance companies
are SOFOM ERs. There are also more
than 1,500 lesser-regulated companies
(SOFOM ENRs in Spanish).
Mexico’s regulated and lesserregulated finance companies are collectively known as SOFOMES. As the
country’s financial system continues
to develop and its commercial banks
cater to more traditional established
markets, regulators view SOFOMES
as a means of addressing basic credit
needs while providing an important
source of liquidity to chronically underserved markets.
SOFOMES are not allowed to accept deposits. Instead, they raise capital from banks, government-sponsored
financial corporations, venture capital
entities or through debt issuance. To

Mexican SOFOM ER Assets, Lending Grow

Index, December 2013 = 100

Share as percent of commercial bank loans

260

240

SOFOM ER loan market share

220

SOFOM ER assets

200

180

160

140

120

100

0
Dec. 2013

Dec. 2014

Dec. 2015

Dec. 2016

Aug. 2017*

*Represents partial-year activity.
NOTE: Total loans are calculated net of loan loss reserves.
SOURCES: Mexican National Banking and Securities Commission (Comisión Nacional Bancaria y de Valores);
authors’ calculations.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

comply with money laundering rules,
SOFOMES must register with Mexico’s
banking and securities commission,
the national financial consumer watchdog agency, the central bank and the
finance ministry. They must comply
with anti-money-laundering measures
and report borrower credit profiles to a
private credit bureau.
SOFOM ERs are subject to
relatively light capital requirements
and accounting standards.3 Mexico’s
commercial banks must maintain a
capital adequacy ratio (total capital to
risk-weighted assets) of 10.5 percent,
while SOFOM ERs must maintain an 8
percent ratio. Moreover, SOFOM ERs
must follow additional regulatory requirements as a result of linkages with
other regulated financial institutions,
such as commercial banks’ community
financing firms.4

Filling Lending Gaps
Mexico is Latin America’s secondlargest economy, bolstered by strong
fundamentals, reform initiatives and
stable institutions.5 Still, financial system development remains a challenge,
especially financial inclusion. Credit
to the private sector and households’
use of deposit accounts are low, new
loan origination by commercial banks
has remained sluggish, and long-term
financing is scarce (Chart 2). Instead of
extending new lines of credit, com-

Chart

mercial banks often consolidate their
financial service offerings, choosing
to improve their existing infrastructure and make larger loans to existing
customers. As a result, a majority of
Mexican adults don’t use the country’s
financial system. The unbanked proportion is greatest in rural areas, where
71 percent of working-age residents
lack access to formal financial services,
far exceeding Mexico’s Latin American
peers (Chart 3).6
Micro-, small- and medium-sized
enterprises also struggle to obtain
credit. A study by the National Statistics
Institute found that only 11 percent of
microenterprises (those with up to 10
employees) have access to financing,
compared with 28 percent for small
enterprises (11–49 employees) and
40 percent for medium-sized entities
(50–250 employees).7 Moreover, when
receiving offers of formal financing,
67 percent of all these businesses turn
them down primarily because of high
borrowing costs, including lenders’
fees, and minimum balance and collateral requirements.
Many of these businesses and
individuals lack established credit histories and operate in Mexico’s large informal sector—the part of the economy
where activity is not reported to the
government and whose participants do
not pay employment taxes or receive
government-mandated benefits and

pensions. The informal sector accounts
for a quarter of Mexico’s gross domestic product, according to the National
Statistics Institute.8
The inability to access credit
through a bank has considerable
economic consequences. Credit-constrained individuals and firms often
can’t take advantage of growth opportunities or absorb financial shocks.
Families are unable to invest in education and health; businesses struggle to
expand and create jobs.
The SOFOMES offer an alternative
borrowing channel. Specifically, because they do not face the same restrictions as banks, SOFOMES can more
easily serve marginalized consumers
without imposing credit, balance and
collateral requirements. SOFOMES undertake higher-risk lending by charging
higher loan interest rates. The average
interest rate on a personal loan taken
at a Mexican commercial bank in April
2017 was 31 percent while at a SOFOM
ER, it was 39 percent. Nevertheless, the
benefits of having access to high-interest financing often outweigh the costs
of being denied credit by a commercial
bank.
Mexico hopes that by striking a
balance between formalization and
innovation within the SOFOME sector,
it can bolster credit, increase investor confidence and encourage new
business formation while discouraging

Loan Issuances and Deposit Volume in Mexico Lag Latin American Peers

Outstanding Loans with Commercial Banks

Outstanding Deposits with Commercial Banks

Percent of gross domestic product

160

160
140

117

120
100

80
34

60
40
20

nt
in
a
M
ex
ico
Ur
ug
Gu uay
at
em
ala
Pe
r
Ec u
ua
do
r
Br
az
Ho
il
nd
ur
Co
a
st s
aR
ica
Be
liz
e
Ch
il
Pa e
na
m
a

Pe
ru
ua
do
r
Br
Gu azil
at
em
ala
Ch
i
Ur le
ug
u
Ho ay
nd
ur
Co
a
st s
aR
ica
Be
liz
Pa e
na
m
a

in
a
ex
ico

100

120

Ar
ge
nt

134

140

SOURCE: International Monetary Fund Financial Access Survey—2017 Edition.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Chart

Financial Inclusion in Mexico Trails Latin America

Percent

60
Mexico
50

hopes that
}Mexico
by striking a balance
between formalization
and innovation within
the SOFOME sector,
it can bolster credit,
increase investor
confidence and
encourage
new business
formation.

Latin America

40
30
20
10
0
Account
ownership

Account
ownership,
rural

Borrowed from
a financial
institution

Credit card
used in the
past year

Domestic
credit to
private sector

NOTE: Percent refers to share of respondents except for domestic credit, where it refers to share of gross domestic
product.
SOURCES: World Bank Global Findex 2014 and International Monetary Fund; International Financial Statistics and data
files; World Bank and Organization for Economic Cooperation and Development.

reliance on costlier and less reliable,
unregulated alternatives.9

Balancing Regulation, Innovation
Mexico has struggled to strike an
optimal regulatory balance. The predecessors to SOFOMES, known as SOFOLES, first appeared in the mid-1990s
after a sharp peso devaluation and
political instability spawned the Tequila
Crisis of 1994—both events disrupting
banking activity. Banks struggled to
rebuild their balance sheets, and the
Mexican government sought to stimulate the credit market.
Nondeposit-taking finance company lending was authorized in the
housing, consumer, small-business and
automobile finance markets.10 Initially,
the finance companies, backed by the
federal government, issued loans for
low-income housing and real estate
development.11
By the mid-2000s, nonbank lending
was common, accounting for nearly half
of Mexican mortgage loan originations.
Commercial banks, operating in a less
stringent regulatory environment before
the 2007–08 global financial crisis, were
drawn to the finance companies and
their mortgage business with the informal workforce.
As the financial crisis unfolded,

Debit card
used in the
past year

delinquency rates soared and SOFOLES
struggled to maintain operations. Total
loans issued by SOFOLES fell 69 percent
between their peak in September 2007
and December 2009, drying up credit
available to individuals and small- and
medium-sized enterprises (Chart 4).
The stress spread to banks that had
purchased finance companies before
the crisis. The downturn exposed fraudulent practices, loose lending standards
and inadequate servicing procedures in
the nonbank financial sector.12
The SOFOLES’ struggles in the
wake of the crisis prompted regulatory
change. Registration of the finance companies was required under laws passed
in 2013, under which SOFOLES were
required to convert into SOFOM ERs or
SOFOM ENRs or dissolve. This led to the
consolidation in the SOFOMES sector.
Regulators also adjusted their oversight
strategy, boosting protection for consumers, mitigating SOFOMES’ lending
risks and scrutinizing the firms for fraud
and money laundering.

SOFOMES in the Future
SOFOMES’ future growth may
come via financial technology (fintech)
companies, which leverage online,
mobile and information technologies
to deliver financial services. Fintechs in

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

Chart

SOFOLES’ Lending Tumbles Following Global Financial Crisis

Billions of pesos

200
180

Loans issued by SOFOLES

160
140
120
100
80
60
40
20
0
Sept. 2007

Sept. 2008

Sept. 2009

Sept. 2010

Sept. 2011

Sept. 2012

SOURCES: Mexican National Banking and Securities Commission (Comisión Nacional Bancaria y de Valores);
authors’ calculations.

developing countries such as Mexico
have objectives similar to SOFOMES,
making financial products, including
loans, available to underserved markets.
To meet this goal, they rely on mobile
device-based transactions and data
analytics.
While regulatory constraints limit
banks’ adoption of these technologies,
fintechs can register and become
SOFOMES. Some existing SOFOMES
have acquired equity stakes in fintech
startups or have developed fintech business lines, technological tools, mobile
apps and other electronic products; others are collaborating with fintech firms.13
The government’s approach to
fintechs appears in line with that of
SOFOMES; regulators seek to balance
innovation with oversight to allow
growth and monitor for fraud.
Still, a large proportion of Mexico’s
population remains without access
to credit, and the SOFOMES, perhaps
with the fintech sector, may increase
financial inclusion.

Perez and Reichow are financial
industry analysts in the Financial
Industry Studies Department at the
Federal Reserve Bank of Dallas.
Notes
SOFOM stands for sociedad financiera de objeto
múltiple. ER stands for entidad regulada. ENR stands
1

for entidad no regulada. SOFOLES stands for sociedad
financiera de objeto limitado.
2
Based on Comisión Nacional Bancaria y de Valores
(Mexican National Banking and Securities Commission)
data, authors’ calculations.
3
See “SOFOMES” by Comisión Nacional Bancaria y
de Valores (Mexican National Banking and Securities
Commission), www.cnbv.gob.mx/SECTORESSUPERVISADOS/OTROS-SUPERVISADOS/
Descripci%C3%B3n-del-Sector/Paginas/SOFOMESReguladas.aspx.
4
See “Padrón de Entidades Supervisadas” (Census
of Supervised Entities), Comisión Nacional Bancaria
y de Valores (Mexican National Banking and
Securities Commission), www.cnbv.gob.mx/Paginas/
PADR%C3%93N-DE-ENTIDADES-SUPERVISADAS.aspx.
5
See “Financial System Stability Assessment” by
International Monetary Fund, November 2016.
6
Working-age individuals are defined as those 15 years
of age or older.
7
See “Se Difunden Estadísticas Detalladas Sobre las
Micro, Pequeñas y Medianas Empresas del País”
(“Statistics on Micro-, Small- and Medium-Sized
Businesses”) by Instituto Nacional de Estadística y
Geografía (National Statistics Institute), Instituto Nacional
del Emprendedor (National Startup Institute) and Banco
Nacional de Comercio Exterior (National Bank of Exterior
Commerce), July 2016, www.inegi.org.mx/saladeprensa/
boletines/2016/especiales/especiales2016_07_02.pdf.
8
See “Actualización de la Medición de la Economía
Informal” (“Actualization in Measuring the Informal
Economy”) Instituto Nacional de Estadística y Geografía
(National Statistics Institute), December 2016, www.
inegi.org.mx/saladeprensa/boletines/2016/especiales/
especiales2016_12_08.pdf.
9
See “A Study on the Effect of Financial Inclusion on the
Relationship Between Income Inequality and Economic
Growth,” by Jong-Hee Kim, Emerging Markets Finance &

Trade, vol. 52, no. 2, 2016, pp. 498–512, www.tandfonline.
com/doi/full/10.1080/154046X.2016.1110467.
10
See “The SOFOLES: Niche Lending or New Leaders
in the Mexican Mortgage Market?” by Natalie Pickering,
Harvard University Joint Center for Housing Studies, May
2000, www.jchs.harvard.edu/sites/jchs.harvard.edu/files/
pickering_w00-2.pdf.
11
SOFOLES typically issued credit for properties between
$14,000 and $22,000 (U.S.). See “The Home Truths About
Non-Bank Mortgage Lending in Mexico,” by Knowledge@
Wharton, The Wharton School, University of Pennsylvania,
Oct. 5, 2011, http://knowledge.wharton.upenn.edu/article/
the-home-truths-about-non-bank-mortgage-lending-inmexico/.
12
See “The Non-Bank Credit Crunch in Mexico: Rise and
Fall of an Industry,” by José Berrospide, Renata Herrerías,
Fabrizio López Gallo and Ana Mier y Terán, Instituto
Tecnológico Autónomo de México (Autonomous Technical
Institute of Mexico), December 2012, http://daac.itam.
mx/sites/default/files/nonbank_credit_crunch_mexico_
dec2012.pdf.
13
See “Es Importante Que las SOFOMES se Sumen al
Huracán Fintech: Banxico” (“It’s Important Sofomes Join
the Fintech Hurricane: Banxico”), by Fernando Gutiérrez,
El Economista, Sept. 10, 2017, www.eleconomista.com.
mx/sectorfinanciero/Es-importante-que-las-sofomes-sesumen-al-huracan-fintech-Banxico-20170910-0043.html.

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

SPOTLIGHT

Rising Education Helps Explain Hispanic
Household Income Growth in Texas
By Alexander T. Abraham and Amy Jordan

ispanic household income
has grown considerably in real
(inflation-adjusted) terms in
Texas and the U.S. during recent years. Household income
is calculated by adding each household
member’s total income from all sources,
including wages, public and private
pension benefits, and financial assets.
Hispanics’ real median household
income grew 13 percent in Texas from
2011 to 2016, compared with 11 percent
nationally and 7 percent for households overall in the state (Chart 1).
Hispanic households make up almost
one-third of all Texas households.
Despite rapid growth, the Texas
Hispanic median household income of
$44,579 in 2016 trails the state median
income of $56,565. The gap is narrower
nationally; the U.S. Hispanic median
household income is $46,882 compared with $57,617 for the U.S. overall.
In Texas, the share of Hispanics
who are college educated is lower than
it is for the total adult population. Nearly
20 percent of all people age 25 and over
hold college degrees; the attainment
rate drops to 10 percent for Hispanics.
Moreover, many Hispanics are immigrants whose wages are lower due to
factors such as limited English proficiency and lack of legal status. Median
household income rises to $49,900 for
Hispanic households headed by a U.S.
native, from $38,580 when the head of
household is foreign born.

Educational Attainment Gains
Greater education is likely driving
the Hispanic income increases. The
share of Hispanics age 25 and older
with a high school diploma increased
1.6 percentage points to 28 percent
from 2011 to 2016, while the share with
bachelor’s degrees rose 1.5 percentage
points to 10 percent. The share without
a high school diploma dropped 4.5
percentage points to 35 percent.

Chart

Hispanic Household Income Rapidly Rises in Texas, U.S.

Index, 2011 = 100

114
112
110
108

All Texas
Texas Hispanics

106
104

U.S. Hispanics

102

All U.S.

100
98
2011

2012

2013

2014

2015

2016

NOTE: Data are inflation adjusted to 2016 dollars.
SOURCE: Census Bureau, American Community Survey.

At the same time, the Hispanic
dropout rate reached a national low.
This development may partly reflect a
shift in the composition of the population, with slowing immigration of Mexican citizens—individuals less likely to
have a high school diploma than other
Hispanic immigrants—and some emigration back to Mexico.1
Between 2005 and 2010, about 1.4
million Mexican immigrants and their
children (including some from Texas) returned to Mexico. The U.S. Mexican-born
population stopped growing in 2007 at
the onset of the Great Recession and
amid tighter immigration enforcement.2
At the same time, violence in
northern Mexico likely contributed to
an influx of relatively highly educated
Mexicans into Texas, although the extent of that migration is unknown.

Economic Conditions Also Improve
The shale oil boom also supported
improved economic conditions for
Hispanics.3 The Hispanic population in
Texas, which stands at nearly 11 million, rose 11 percent in 2011–16. The oil

boom resulted in more blue-collar jobs
in high-paying energy and manufacturing sectors. Hispanic wages, which in
Texas accounted for slightly more than
75 percent of average Hispanic household income in 2016, spiked 6.5 percent
around the time of the boom.
While Hispanic households have
made economic advances, income
inequality remains a concern. A greater
share of Hispanics live below the poverty line (22 percent) than the overall
share of the state population in poverty
(16 percent). Furthermore, 27 percent
of Hispanics in Texas lack health insurance; the overall state uninsured rate is
17 percent.
Notes
See “A Look at Immigrant Youth: Prospects and
Promising Practices,” by Ann Morse, National
Conference of State Legislatures Children’s Policy
Initiative, March 2005.
2
See “Net Migration from Mexico Falls to Zero—and
Perhaps Less,” by Jeffrey Passel, D’Vera Cohn and Ana
Gonzalez-Barrera, Pew Hispanic Center, April 23, 2012.
3
See “The Texas Energy Industry: From Boom to Gloom,”
by Michael D. Plante and Mine K. Yücel, Federal Reserve
Bank of Dallas Annual Report 2015.
1

Southwest Economy • Federal Reserve Bank of Dallas • Fourth Quarter 2017

GO FIGURE

Remittances to Central America Soar
Design: Emily Rogers & Darcy Taj; Content: Stephanie Gullo & Jesus Cañas

Money Sent to Central America on the Rise
Total remittances received, index, 2000 = 100
450
400
350
300
250
200
150
100

Great
Recession

... but 1.6 times more remittances were sent to
Mexico than Central America in 2016.
Central America $18.3 billion

Mexico $28.7 billion

‘00 ‘01 ‘02 ‘03 ‘04 ‘05 ‘06 ‘07 ‘08 ‘09 ‘10 ‘11 ‘12 ‘13 ‘14 ‘15 ‘16

What’s Behind the Change?
Central American immigrant population in U.S. increasing,
while Mexican immigrant population in U.S. decreasing ...

Central Americans in U.S. send more
money home per person on average

Population in U.S., index, 2010 = 100
115

$325

110

$225

105
100
95
‘10

‘11

‘12

‘13

‘14

‘15

‘16

... but Central American population still far smaller
than Mexican immigrant population

What Does This Mean
for Receiving Countries?

Central America

Mexico

3.5 million
11.6 million

Remittances as Percent of GDP

Mexico
Remittances provide relatively bigger boost
to Central American economies, which are
more dependent on them than Mexico.

2.7%

Belize—4.9%

Remittances Per Capita in Receiving Countries
El Salvador—$724

Belize—$238

Panama—$125

Guatemala—$450

Mexico—$225

Costa Rica—$113

Honduras—$422

Nicaragua—$206

NOTES: Dollar values are inflation-adjusted 2016 dollars. Central America includes Belize, Costa Rica, El Salvador,
Guatemala, Honduras, Nicaragua and Panama. All data refer to 2016 unless otherwise noted.
SOURCES: Census Bureau, American Community Survey, 1-year estimates; World Bank.

Honduras—17.9%
Guatemala—10.9%

Nicaragua—9.6%

El Salvador—17.1%
Costa Rica—1.0%

Central America

7.5%

Panama—0.9%

Texas Sees Coverage Gains Under Health Care Act
(Continued from page 7)
6
See “Medicare Income Eligibility Limits for Parents,
2002–2017,” Kaiser Family Foundation, www.kff.org/
medicaid/state-indicator/medicaid-income-eligibilitylimits-for-parents.
7
Part of this decline could be due to changes in how
the eligibility limit is calculated post-ACA. The cutoff
published by the Texas Health and Human Services
Commission differs slightly—$230 in monthly income
or 16 percent of the FPL.
8
Enrollment numbers are based on Medicaid data for
July 2017. See www.medicaid.gov/medicaid/programinformation/medicaid-and-chip-enrollment-data/reporthighlights/index.html. Medicaid caseload data from the
Texas Health and Human Services Commission indicate
that Texas’ Medicaid enrollment rose 13.4 percent in 2014.
9
See “2017 Marketplace Plan Selections with Finance
Assistance,” Henry J. Kaiser Family Foundation, 2017.
10
The analysis of 27–64-year-olds with no college
education is based on CPS-IPUMS data. See Integrated

DALLASFED

Public Use Microdata Series, Current Population Survey:
Version 4.0 [dataset], by Sarah Flood, Miriam King,
Steven Ruggles and J. Robert Warren, University of
Minnesota, 2015, http://doi.org/10.18128/D030.V4.0.
11
See Federal Subsidies for Health Insurance Coverage
for People Under Age 65: Tables From CBO’s March
2016 Baseline, Congressional Budget Office, www.cbo.
gov/sites/default/files/recurringdata/51298-2016-03healthinsurance.pdf.
12
See “How CBO Estimates the Effects of the Affordable
Care Act on the Labor Market,” by Edward Harris and
Shannon Mok, Congressional Budget Office, Working
Paper no. 2015-09, December 2015.
13
See “The Effects of the Affordable Care Act on Health
Insurance Coverage and Labor Market Outcomes,” by
Mark Duggan, Gopi Shah Goda and Emilie Jackson,
National Bureau of Economic Research, NBER Working
Paper no. 23607, July 2017.
14
See “Health Insurance and the Consumer Bankruptcy

Southwest Economy

is published quarterly by the Federal Reserve Bank of
Dallas. The views expressed are those of the authors and
should not be attributed to the Federal Reserve Bank of
Dallas or the Federal Reserve System.
Articles may be reprinted on the condition that the
source is credited to the Federal Reserve Bank of Dallas.
Southwest Economy is available on the Dallas Fed
website, www.dallasfed.org.

Federal Reserve Bank of Dallas
2200 N. Pearl St., Dallas, TX 75201

Decision: Evidence from Expansions of Medicaid,” by Tal
Gross and Matthew J. Notowidigdo, Journal of Public
Economics, vol. 95, no. 7, 2011, pp. 767–78.
15
See “The Impact of Medicaid Expansion on
Uncompensated Care Costs,” by Deborah Bachrach,
Patricia Boozang and Mindy Lipson, Robert Wood
Johnson Foundation, June 2015, www.rwjf.org/en/
library/research/2015/06/the-impact-of-medicaidexpansion-on-uncompensated-care-costs.html.
16
See, for example “Public Health Insurance and Private
Savings,” by Jonathan Gruber and Aaron Yelowitz,
Journal of Political Economy, vol. 107, no. 6, 1999, pp.
1,249–74.
17
See “The Stimulative Effect of Redistribution,” by Bart
Hobijn and Alexander Nussbacher, FRBSF Economic
Letter, Federal Reserve Bank of San Francisco, no.
2015–21, June 2015.

Marc P. Giannoni, Senior Vice President and Director of Research
Pia Orrenius, Keith R. Phillips, Executive Editors
Michael Weiss, Editor
Kathy Thacker, Associate Editor
Dianne Tunnell, Associate Editor
Ellah Piña, Graphic Designer

Full text of The Southwest Economy : Fourth Quarter 2017

FRASER