Macroeconomic, sectoral and distributional effects of

the Infrastructure Investment and Jobs Act in the

United States

David Suarez-Cuesta, Maria C. Latorre, and Robert Z. Lawrence

Abstract
We offer the first analysis of the Infrastructure Investment and Jobs Act (IIJA) using a
Computable General Equilibrium approach, to the best of our knowledge. We use The Enormous
Regional Model (TERM) for the U.S. economy, which allows us to introduce regional-level shocks
and offers macroeconomic results, industry-level results and disaggregated results for ten
different occupations (skilled and unskilled). We evaluate the short run effects of the increase
in government demand for Construction and the long run effects of the stimulus once the
investments rise the country’s physical capital stock. The debt increase caused by the upfront
investment can be financed through foreign debt (i.e., a trade deficit) or through reduced
national savings. We evaluate the impact under both scenarios, considering that this contrast
between national borrowing and foreign borrowing can also be interpreted as a differential
impact between short and long-term debt maturity periods. The accumulation of an important
debt, which has to be paid back in a given point of time in the future, tends to translate into
abrupt production adjustments and even output contraction in some sectors. By contrast, a
more regular and evenly distributed debt service tends to facilitate a strong boost for production
across all sectors and workers categories, with unskilled workers (those without a university
degree) benefiting the most. The plan unleashes a general expansion of wages and employment
across most occupations, with only a few exceptions remaining nearly unaffected. In general,
although most workers categories are better off, unskilled workers benefit more than skilled
ones in terms of the number of jobs created and wages. In addition, labor increases its share
while capital slightly diminishes it in total GDP.

1. Introduction
The dramatic situation that the world has experienced with the Covid-19 pandemic has been
later exacerbated with Russia’s invasion of Ukraine, and many countries have taken different
public policy measures to revitalize their economies. Most of these measures have something
in common, namely, they have increased public spending to deal with the crisis. In November
2021, U.S. President Biden signed the Infrastructure Investment and Jobs Act (IIJA), a law that
aims to “rebuild America’s roads, bridges and rails, expand access to clean drinking water,
ensure every American has access to high-speed internet, tackle the climate crisis, advance
environmental justice, and invest in communities that have too often been left behind” (The
White House, 2021a). The IIJA package includes $550 bn in new spending over 2022 through
2026 to improve the country's infrastructure (United States Congress, 2021), and 650$ bn
coming from existing infrastructure funds and repurposed funds from other areas. It goes
beyond the $400 billion of fiscal incentives created by the U.S. Inflation Reduction Act (The

1
Economist, 2022). The government plans to rebuild America’s infrastructure, “creating good
middle-class union jobs, supporting disadvantaged and underserved communities, advancing
climate resilience and sustainability, and investing in American manufacturers.” (The White
House, 2021b).

This law represents the most ambitious infrastructure investment in the United States (U.S.)
since President Eisenhower created in 1956 the Interstate Highway System, the greatest public
works project in American history (Blas, 2010). According to the White House, it breaks records
in terms of investment in public transit, passenger rail, single dedicated bridge and clean
drinking water and wastewater infrastructures in American history. According to the 2019
Global Competitiveness Report (Schwab, 2019), the US ranks 13th in global infrastructure quality
and 17th in road infrastructure quality. Over the past decades, government infrastructure has
aged dramatically. The American Society of Civil Engineers (2021) classified America’s
Infrastructure as “poor” (C- score), and advocates for an increased, long-term investment to
close the 2.6$ trillion investment gap. According to this professional organization, the
deterioration of transportation infrastructures in the U.S. entails significant direct and indirect
costs for the country's economy. They estimate America’s overdue infrastructure bill will cost
by 2039 the average household $3,300 a year.

Public spending on basic infrastructures1 (which includes capital and “operation & maintenance”
expenses) as a share of GDP has declined since the 1960’s in the U.S. Moreover, over the past
decades the share of total public spending on basic infrastructure devoted to operation &
maintenance has been increasing over the share devoted to public capital spending, which is a
clear signal of infrastructures aging. Figure 1 shows the evolution of annual total public capital
spending (blue line) and the “operations and maintenance” spending (orange line) on basic
infrastructures since 1956 (CBO, 2018). In addition, it shows the share of both of them over GDP
(blue and orange spotted lines, respectively).

In 2002 annual public spending on capital of basic infrastructure peaked at 1.8% of GDP ($243
bn) and decreased to 0.9% of GDP ($174 bn) in 2017. The IIJA’s initial plan is to spread that $550
in the 2022-26 period, which would increase current annual capital infrastructure spending by
$110 bn. With that increase there would be a contained annual increase of 0.55 percentage
points (raising from 0.9% to 1.5%). Although considerably reduced from its first proposal ($2.3
bn), this plan changes the infrastructure investment over GDP declining trend in the US since
the 1970’s decade. With the IIJA, this ratio would still stay under the historical mean but
counteracts the downward trend of the past decades.

1
We follow the definition of “basic infrastructure” given by BEA (Bennett et al., 2019). It includes roads,
railroads, bridges, and rails - https://www.bea.gov/system/files/papers/BEA-WP2020-12.pdf

2
 Figure 1. Total public2 infrastructure spending for capital of basic infrastructures, public
spending for operation and maintenance of basic infrastructures and the public spending
increase due to the IIJA (in $US billion and as a share of GDP)

Source: authors’ own elaboration based on CBO (2018).

The main purpose of this paper is to talk to the American debate and shed light on whether or
not the public investment stimulus will truly reduce the gap between skilled and unskilled labor.
To the best of our knowledge, we offer the first CGE-based analysis on the IIJA, and there are
not many analyses on this law. Some exceptions are the studies by Bonakdarpour et al. (2021)
and Yaros & Zandi (2021), which were undertaken prior to the publication of the IIJA. In this
paper we offer new short- and long-run estimations based on the information available on the
IIJA and on ex-ante Computable General Equilibrium (CGE) simulations, using The Enormous
Regional Model (TERM) for the U.S. Short-term effects are those occurring during the
construction phase, in which the stimulus is based on a government demand increase for
construction. By contrast, in the long-term infrastructures are fully functional and become part
of a country’s physical capital. Our methodology allows us to provide a very thorough analysis
for macroeconomic variables, such as GDP, aggregate private and public consumption,
investment, exports, imports, wages and employment. We also derive the outcomes for
production across the 26 sectors in which we have divided the U.S. economy and pay particular
attention to the distributional impact across 10 types of occupations3 (two skilled and eight
unskilled). Our analysis takes into account different ways of financing the bill, namely, through
foreign or national borrowing. In the real world, probably the funding will be a mix of foreign
and national borrowing, however we want to illustrate their different implications.

This paper is organized as follows. The next section describes the Infrastructure Investment and
Jobs Act (IIJA) and breaks down the various items corresponding to new investment. In section
3 we describe the TERM model used for the simulations considering different ways of financing

2
Federal, state and local.
3
International Standard Classification of Occupations (ISCO-08)

3
the deficit generated by the increase in public spending. Section 4 presents the simulations and
macroeconomic results, while sections 5 and 6 present the impact on production and across
occupations. Finally, the last section draws several conclusions.

2. Overview of the Infrastructure Investment and Jobs Act

Our impact analysis focuses on the $550 bn within the IIJA that correspond to new federal
investment on basic infrastructure. When a $1.2 trillion is offered as the figure related to Biden’s
plan (e.g., U.S. Department of Transportation, 2022) it includes additional funding normally
allocated each year for highways and other infrastructure (e.g., Federal-aid Highway Program,
Federal Railroad programs or Research, Technology & Education). Thus, the funding beyond
$550 billion is already included in the figures of government spending in our model and does
not imply any additional government spending. Aggregate transportation investment includes
private and public investments. The former constitutes the highest share of aggregate
transportation investment and is dedicated to transportation equipment, while public
investment is devoted to transportation infrastructure (U.S. Department of Transportation,
2022). Although the IIJA also includes investment in equipment (e.g., electric vehicles, buses),
most of it is dedicated to construction and renovation of infrastructure, and that is what we are
considering in this study.

Table 1 shows what the new investment funds will be devoted to. The law contemplates more
than $110 billion to repair roads, bridges and highways, and $66 billion to promote passenger
and freight rail. The latter investment represents the largest injection of funds into the country's
rail network in the past five decades. By dedicating $65 billion for broadband infrastructures, it
aims to supply every American with high-speed internet access. The bill also allocates $65 billion
to modernize the electric grid and provides $63.3 for water infrastructure, e.g., lead pipe
replacement. Related to transit it also breaks records by allocating more than $54 billion in
transit modernization, replacement and acquisition of vehicles and for the expansion of the
electric vehicle charging network. It also allocates billions of dollars to increase cybersecurity,
reduce climate change effects, as well as to reform irrigation, ports and airports infrastructures,
among others.
Table 1. New investment allocation of funds (in $US billion)
Additional annual
Model Total spending 2 Annual increase
Category 1 spending 2022-26
sector 2017 ($US billion) 2022-26 (%)
($US billion)
Roads and bridges 93.8 22.0 23%
Passenger and freight rail 4.4 13.2 298%
Transit 21.3 10.8 51%
Transport
Airports 11.4 5.0 44%
Ports and waterways 3.7 3.4 93%
Safety 1.2 2.2 183%
Clean energy and grid 12.8 13.0 102%
Water infrastructures 39.4 12.7 32%
Utilities
Resiliency - 10.0 -
Environmental remediation 1.0 4.2 420%
Information Broadband 6.0 13 217%
Total 195.0 110 56%
Source: authors’ elaboration based on The White House (2021a), CBO (2018), NHSTA (2017),
DOE (2017), EPA (2016), GAO (2020).
Notes:

4
1
Latest data from Congressional Budget Office.
2
Total funds equally distributed from 2022 through 2026.

The infrastructures that are built in the short term imply a short run increase in the demand for
construction. Once they are built, they can be grouped into three main sectors4: Utilities ($181
billion for energy distribution infrastructures, water, sewage services), Transport services ($298
billion for land transport, air transport, water transport, transit) and Information services ($65
billion for telecommunications, cable networks, information services, data processing services).
Thus, the IIJA will increase the capital stock in these three sectors in the long run.
Regarding the legislation pay for, the plan initially proposed offsets that would cover the $550
billion of new spending (Cantwell, 2021). However, recent estimates find that total savings will
cover less than half of it, as some of the reported offsets have already taken place (Committee
for a Responsible Federal Budget, 2021).

3. Methodology and data

We use the United States version of The Enormous Regional Model (TERM-USA), a Computable
General Equilibrium (CGE) model. CGEs are based on mathematical equations that capture the
interactions between factor markets and goods markets and can integrate both the macro and
microeconomic levels (Latorre 2012a, chapter 1; Latorre et al., 2020). They incorporate real data
into a rigorous theoretical framework and present the relationships among economic agents as
a system of equations derived from microeconomic optimization theory and cover the behavior
of households, firms and government in the economy. Resting on the usual progression of the
circular flow of the economy (production, income distribution and domestic and foreign
demand), CGEs describe the equilibrium conditions in goods, factor markets and in the foreign
sector. Moreover, they are flexible tools able to take into account many economic sectors
considering the linkages among them and assess which of them benefit or would be negatively
affected after a policy shock. This methodology is therefore appropriate to analyze the
quantitative impact of the infrastructure bill at both the macro and the micro economic level.

The TERM framework was originally developed for the Australian economy (Horridge, 2012),
and builds on the ORANI model (Dixon et al., 1982). Its database was developed using detailed
official and regional statistics (Horridge, 2012). It follows a bottom-up strategy, meaning that
regional results sum up to obtain the national results. Specific versions of the TERM model have
been developed for a wide variety of impact analysis, such as Australian water shortage / water
trading events (Wittwer & Dixon, 2013), or deforestation in Brazil (Souza Ferreira Filho et al.,
2015). Regarding public infrastructure related studies, Giesecke et al. (2008) analyzed regional
government infrastructure provision under different financing scenarios, and Horridge &
Wittwer (2008) estimated the economic impacts of a construction project using SinoTERM, a
TERM version for China.

The United States version of the TERM used in this study divides the US economy into 11 regions,
26 sectors and 10 occupations. This allows us to offer a rich set of estimates at the sectoral level
and across different types of workers. This granular approach helps to identify whether Biden’s

4
Although some sub-items could be classified in other sectors, these correspond to comparatively much
smaller amounts. Therefore, we do not model them explicitly, which implies we are conservative in our
analysis.

5
act would help to counteract inequality issues that have been identified in the US economy and
for which increases in investment could be positive (Lawrence, forthcoming; Lawrence, 2015).
Moreover, our study would provide a real-world test on whether new government programs
that emphasize construction for infrastructure, renewable energy and climate change mitigation
could serve to offset some of the non-inclusive impacts that have been found in R&D-intensive
programs (Lawrence, forthcoming). We have updated data on GDP and employment to 2019
levels5 using the TERMSCAL program developed by the Center of Policy Studies (2019). See
Appendix 3 for a more detailed description of the model version used in this study.

4. Simulations and macroeconomic results

The economic effects of an infrastructure investment plan are usually approached in two
phases: (1) short-terms effects during construction phase, in which the stimulus is based on a
government demand increase for construction and (2) long-term effects are those during the
operational phase, in which infrastructures are fully functional and increase physical capital in
Utilities, Transport services and Information services sectors. In the long run the increase in
government demand ceases but its effects work through a larger and better infrastructure6. This
study analyzes both phases, estimating the impact after the first year (short run) and in the long
run. We define shocks on a percentage change basis (2019 as base year), using an estimated
distribution of funds across states7 (see Table in Appendix 1).

For the short run simulations, we show results after the first year. We run a government
spending increase in the Construction sector, following the information available on the IIJA.
Considering the $550 billion in new investments are distributed equally between 2022 and 2026,
government spending increases in one year by 2.55% from initial levels ($110 billion each year).
This would represent the highest annual government spending increase in the last decades. In
the short run, the economy’s production response to an external shock is limited, and
investment flows are not yet transformed into capital stocks. Therefore, we keep capital stocks
fixed, while we allow the rental price of capital to vary. In the first year, wages are unable to
vary, whereas employment adjusts to meet the demand increase.

The long run simulations contemplate the $550 billion devoted to new public spending and are
based on three premises: investment stimulus raises capital stocks, output productivity
increases and employment returns to its initial levels. Capital stocks in all industries are now
determined endogenously, except for the Utilities, Transport services and Information services
industries. We exogenously increase the capital stock in these three latter industries, based on
allocation of funds shown in Table 1. Regarding output productivity, a meta study by Bom &
Ligthart (2014) aggregated almost three decades of studies on measuring the private output
elasticity of public capital. They found a long-run output elasticity of public capital supplied at

5
Latest data available (from the U.S. Census Bureau) corresponds to 2020 (Census, 2022). The reader may
check that government spending did not follow the previous trend due to the Covid-19 pandemic
outbreak. Therefore, we consider 2019 to be a more realistic base year for our shock estimations, as data
for full-year 2021 is unavailable at the time of running the simulations (March 2022).
6
The Bureau of Transportation Statistics defines capital expenditure as any expenditure that adds to the
productive capacity of the economy, i.e., includes construction and
reconstruction/rehabilitation/restoration of existing infrastructures except for routine maintenance
expenditures.
7
Calculations are set by statute and can vary over time. Estimates are based on prior legislation that could
vary over time.

6
the central government level of 0.122. Based on this finding, we apply that elasticity to increase
productivity in Transportation, Utilities and Information considering capital stock increase in
each industry. Differently from the short run closure rules, in the long run we assume
employment returns to initial levels, while national real wages adjust to labor market conditions.

When government spending leads to a larger federal budget deficit, it usually reduces the
national savings rate and raises the trade deficit. A portion of the budget deficit is effectively
financed through a rise in the total amount Americans borrow from abroad (McBride & Chatzky,
2019). Based on this, we define a scenario in which the initial stimulus is financed by the foreign
sector (international borrowing) and a second scenario in which the increase in public spending
is financed through reduced national savings (national borrowing). The investment stimulus is
not financed through higher taxes in any of the simulations.

Because the labor market serves as primary driver of much of the activity in each region, in all
scenarios we make regional consumption follow regional wage income. Table 2 shows the short
and long run results for the main macroeconomic variables under each scenario. The real impact
(in both the short- and long-run) would be somewhere in the middle of the national or foreign
scenarios we analyze since it will probably be a mix of national and foreign borrowing. All the
macroeconomic variables in Table 2 are expressed in real terms and in percentage change with
respect to the initial data.

Table 2. Impact on macroeconomic variables

(% change with respect to initial levels)

Foreign borrowing National borrowing

Macro variables
Short run Long run Short run Long run
Private consumption 0.00 0.00 -0.17 1.44
Investment 0.00 0.00 -0.35 0.56
Government consumption 2.55 0.01 2.55 1.46
Exports -0.72 6.61 0.02 1.35
Imports 0.61 -2.45 0.12 0.70
Real GDP 0.22 1.18 0.19 1.39
Agg. Employment 0.43 0.00 0.38 0.00
Real wage 0.00 2.76 0.00 3.94
Agg. Capital Stock 0.00 2.10 0.00 2.52
GDP Price Index 0.65 -3.90 0.12 -0.64
Export Price Index 0.36 -3.15 -0.01 -0.67
Source: authors’ estimations.
Notes: results are expressed in real percentage changes from initial levels.

Under (1) foreign borrowing (see Table 2), the increase in government spending will be financed
by the foreign sector, i.e., through a trade deficit increase. Thus, in addition to the short- and
long- run assumptions already mentioned, we fix private consumption and investment. In the
short run, the rise in demand for construction output increases demand for labor, and aggregate
employment rises by 0.43% (equivalent to 640,000 workers). As a result, GDP goes up by 0.22%.
The investment stimulus causes a rise in the domestic price of goods and services, and the GDP
price index and export prices increase by 0.65% and 0.36%, respectively. This domestic
appreciation causes a domestic increase in demand followed by a fall in exports (0.72%) and a

7
rise in imports (0.61%), and the trade deficit rises by $35.1 billion after the first year. A similar
short run macroeconomic adjustment was experienced during the Global Financial Crisis, in
which the domestic increase of construction investment generated a negative effect on the
trade balance. This was reinforced by the increase in domestic construction prices which, in turn,
caused a currency appreciation. However, dwelling investment is different than infrastructure
investment, and the latter should boost production in the economy more heavily than the
former. Improvements in basic infrastructure help to reduce firms’ costs, and to boost
productivity in the economy in the long run.

In fact, the competitiveness loss we have just described for the short run is reversed in the long
run, when the initial deficit increase will have to be financed (i.e., the foreign debt has to be paid
back) and infrastructure investments increase capital stocks. Thus, in the long run imports
decrease by 2.45% and exports increase by 6.61%, leaving an overall trade deficit reduction of
$41.7 billion from initial levels.

In the long run, the increase in capital stocks and in output productivity rises the productive
capacity of the economy, and GDP rises by 1.18%. Workers experience an increase in the capital
available for production and become, thus, more productive. As a result, their wages go up by
2.76%.

Under (2) national borrowing (see Table 2) the increase in government spending in
infrastructure is not financed through higher taxes, but through a reduction in private
consumption and investment8. Thus, we allow private consumption and investment to vary
while we fix the balance of trade as a share of GDP, assuming that the national deficit increase
is financed internally through a redirection of savings or waiving current consumption levels. In
the short run, the economy grows similarly to the foreign borrowing scenario, but the deficit
increase caused by the initial stimulus is financed through reduced private consumption (0.17%)
and reduced investment (0.35%), mainly because the extra government spending (increase in
aggregate demand) puts upward pressure on prices (the fixed trade balance avoids changes in
foreign savings). GDP rises by 0.19% and aggregate employment by 0.38%. As in the foreign
borrowing scenario, in the long run the increase in capital stocks and output productivity rises
the productive capacity of the economy. GDP and national real wages increase by 1.39% and
3.94%, respectively. Do note, however, that both the increase in GDP and wages are now
somewhat larger than in the foreign borrowing case.

With national borrowing the increase in government spending is financed simultaneously in the
short run through reduced private consumption and investment. This implies that in the long
run, there is no debt (and no interest from the debt) to pay back, contrary to the case of foreign
borrowing. As a result, in the long run under national borrowing the expansionary trend caused
by a larger capital stock and productivity increases, boosts consumption and investment.
Therefore, the expansionary trend is larger than with foreign borrowing. This contrasts with the
impact in the short run, in which the sacrifices in consumption and investment necessary to
nationally finance the increase in government spending slightly reduce the positive impact of
foreign borrowing. Do note however, that even though we only present the results in the short
run for one year, this shock represents one fifth of the overall Biden’s stimulus package which

8
The model does not differentiate between private and public savings, nor does it include interest rates,
which is common in models using this methodology. The higher prices are act like a mechanism similar to
increases in interest rates.

8
will be implemented in five subsequent years. Thus, the results we present for one year could
more or less be multiplied by five, to obtain an insight into their cumulative impact as the IIJA
unfolds in a 5-year period.

Under national borrowing no foreign borrowing occurs. Therefore, we consider a constant

balance of trade over GDP ratio, and thus exports and imports adjust to meet this constraint
(exports grow by 1.35% and imports grow by 0.70%). The counterpart of this increase in
competitiveness is a decrease in the general price of goods and services, i.e., the economy
experiences a deflationary process in the long run (GDP price index decreases by 0.64%).

We compare our results with two relevant analyses that have been conducted on the IIJA,
namely, Bonakdarpour et al. (2021) and Yaros & Zandi (2021). Both were published prior to the
approval of the law and use an econometric approach. To our knowledge, our impact analysis is
the first study of the IIJA that uses a CGE model. Moreover, our results are more comprehensive,
as we provide results for several macro variables which the rest of the analyses (to date) do not
offer. In particular, their studies do not cover the impact on exports, imports and prices.

Although macro-econometric models and CGEs are both based on a core input-output structure
and on a national accounting framework, they offer a different approach to phenomena. Macro-
econometric models do not include optimizing behavior as they are based on uncertainty
conditions (Pollitt et al., 2019; Latorre et al., 2020). Unlike econometric models, in which
parameters are based on time-series data, CGE’s are based on microeconomic behavioral
equations that capture the links between the different agents in an economy.

The study carried out by Yaros & Zandi (2021) is mentioned in The Annual Report of the Council
of Economic Advisers (United States President & Council of Economic Advisers, 2022). It
compares a business as usual scenario with several possible scenarios, including the impact of
the America Rescue Plan9 (ARP) (The White House, 2021c), the IIJA corresponding to additional
infrastructure investment ($550 billion) and the Build Back Better framework. Differently from
our analysis, they do not isolate the effect of the IIJA, and make their estimates based on an
increasing distribution of investment between 2022 and 2031. Results are shown as an add-on
to the ARP, comparing a scenario in which only ARP is contemplated with a scenario in which
ARP and IIJA are combined. Therefore, comparations should be taken cautiously. The IIJA could
rise US GDP by 0.6% in 2023, if we consider the difference between the impact of ARP + IIJA
(2.9%) and the impact of the ARP alone (2.3%). However, in the following years the impact of
the IIJA seems to be close to zero (since it coincides with the impact of ARP + IIJA) or even
negative, with a tenth of a percentage point difference (i.e., 1.7% with IIJA vs 1.8% without IIJA).
As for employment, they estimate an increase of 320,000 workers10 for the same year, reaching
800,000 by 2025. In terms of jobs created, their results point to fewer jobs than we estimate. A
larger GDP with smaller job creation than we find may occur for example due to the sectors in
which jobs are created and for which they offer no detail. This study uses The Moody’s Analytics
macro-econometric model for the simulations over the decade through 2031.

Bonakdarpour et al. (2021) study was commissioned by the American Road and Transportation
Builders Association and analyzes the additional highway, bridge and public transit spending in
the IIJA (around $150 billion). It uses a combination of two models, the IHS Markit model and

9
Signed into law by President Biden in March 2021.
10
They show level-form results for the number of jobs created, while our results are expressed in
percentage change. Therefore, comparisons should be made with caution.

9
the IMPLAN model. Considering that these results show the dynamic impact of $150 billion
on the U.S. economy11 (i.e., about a third of the shock we run) their GDP estimates are of a
total accumulation of $488 billion between 2022 and 2027, as well as an employment
increase of 200,000 workers per year in the same period. Thus, as happened with Yaros &
Zandi (2021), Bonakdarpour et al. (2021) also obtain larger GDP estimations but lower job
creation than we find.
Other major public investment projects have been analyzed under a CGE framework (Euijune et
al., 2004; Zhai, 2018; Christensen et al., 2019). The RHOMOLO is the spatial computable general
equilibrium model of the European Commission used to analyze the impact of the Investment
Plan for Europe, also called the Juncker Plan (European Commission, 2014). Christensen et al.
(2019) uses the RHOMOLO-EIB12 model, which analyses the effect of the EU structural
investment funds provided through lending. According to their study the total expected
investments from the European Fund for Strategic Investments (under the Juncker Plan) totaled
€408 billion over the 2015-2019 period. Their results showed that these investments would
cause a 1.76% increase on GDP by 2022, as well as an increase of 1.68 million workers by the
same year. A smaller amount of investment, accumulated through a similar period of time (5
years) yields a somewhat larger GDP impact and smaller job creation than we find. However,
comparations should be made cautiously, as these results are referred to a different region (EU).

5. Impact on production
Table 3 presents the impact on output at a sectoral level for each scenario in the short- and long-
run. Short run estimations are based on an increase in public demand for construction output,
and therefore this sector experiences the highest increases. A high share of the goods produced
by some industries such as Non-metallic mineral products (37%), Other mining (21.8%) or Metal
products (11.4%) is used as intermediates by the Construction sector, and therefore these
industries experience important output increases. The Construction sector performs slightly
better in the short run when the investment stimulus is financed through an increase in the
trade deficit (foreign borrowing). This is because under national borrowing investment goes
down, thus reducing slightly private demand for construction. Production in the rest of the
sectors remains nearly unaffected in the short run.

In the long run the sectors targeted by the IIJA experience larger increases in the capital stock
than the rest. This explains why Utilities, in particular, and to a lesser extent Transport services
and Information services exhibit quite remarkable increases in production. Although the
Transport services sector receives most of the investment, it also experiences smaller output
increases than the other two. This is due to the fact that this sector is the one that increases
comparatively less its capital stock (in percentage terms). The increase in demand for Utilities
pulls demand for Energy minerals products, which also experiences an important increase in
production. More than 37% of the Energy and Minerals sectors’ output is used as intermediate

11
Sum of investment on highways & bridges ($110 bn) and public transit ($39.4 bn).
12
The RHOMOLO-EIB framework capitalizes on the well-established RHOMOLO model, initially
developed by the Joint Research Centre to evaluate the performance of EU policies and extends it to
cover the business model of the European Investment Bank (EIB) Group. The main difference between
the two versions of the model lies in the modelling of the EIB-Group operations as loans financed by the
Bank which differ from grants financed by taxes levied by the EU governments. The latter are the ones
normally analyzed with the RHOMOLO model.

10
by the Utilities sector, and almost 45% of the intermediates used by the Utilities industry comes
from the Energy minerals industry. In addition, because there is a general increase in the U.S.
capital stock in the long run there is an overall expansionary trend in most of the sectors of the
economy. This contrasts with their sluggishness in the short run, with the exception of
construction and its suppliers, which grow in the short run.
Table 3. Impact on sectoral production
(% change with respect to initial levels)

Foreign borrowing National borrowing

Sector
Short run Long run Short run Long run
Total Agriculture -0.03 1.09 0.01 0.87
Crops -0.03 1.03 -0.01 0.95
Livestock -0.02 0.35 -0.03 0.99
Forestry & Fishery products -0.01 1.64 0.10 0.45
Total Manufacturing 0.93 2.09 0.91 1.56
Energy minerals -0.01 6.24 0.02 4.34
Other mining 0.44 2.61 0.48 1.28
Utilities 0.01 8.77 -0.01 9.08
Construction 3.66 -0.74 3.34 0.37
Food, Drinks, Tobacco -0.09 0.49 -0.11 0.97
Fabrics, textiles, clothing -0.18 1.48 -0.09 1.20
Wood products, paper 0.15 1.38 0.22 0.80
Petroleum & Chemicals 0.00 1.78 0.04 1.11
Rubber & Plastic 0.13 1.73 0.23 0.73
Non-metallic mineral products n.e.c. 0.84 1.19 0.85 0.96
Metal products 0.16 2.77 0.35 0.72
Machinery & Equipment n.e.c. -0.08 2.83 0.15 0.50
Transportation equipment 0.10 1.57 0.21 0.34
Miscellaneous manufacturing -0.10 1.43 -0.04 0.66
Total Services 0.03 0.74 0.01 1.14
Trade services 0.13 0.49 0.04 0.90
Transport services 0.07 2.77 0.10 2.36
Information services 0.04 5.18 0.00 5.19
Business & Professional services 0.06 1.09 0.07 0.74
Education -0.02 -0.18 -0.15 0.58
Health & Social services -0.01 -0.50 -0.16 0.35
Other services -0.07 1.34 -0.04 1.06
Government & Defense 0.00 -0.08 0.00 1.23
Owner-occupied dwellings 0.00 -0.61 0.00 0.74
Total 0.22 1.18 0.19 1.39
Source: authors’ elaboration.
Notes: results are expressed in percentage change from initial levels.

Another contrast stands out in the evolution of sectoral production. As we say, an overall
expansionary trend in production is clear both under foreign borrowing and national borrowing
in the long run. However, consistently with a somewhat smaller increase in the capital stock
under foreign borrowing than with national borrowing (Table 2), we see that sectors like
construction and a few services sectors experience a reduction in production only under foreign
borrowing. In this scenario the debt incurred is paid back in the long run, with private
consumption and investment remaining fixed in a context in which the GDP is growing (Table 2).

11
The latter implies a smaller push for national demand, since national savings are increasing to
pay back the debt, which translates into a low investment in construction and small demand for
a few services sectors. As a result, these latter sectors reduce production. By contrast, with
national borrowing the increase in government spending is simultaneously financed through
reduced private consumption and investment, as the short run adjustment in Table 2 shows.
Therefore, in the long run (under national borrowing), there is no need to pay back the debt. In
fact, the debt has already been cancelled year by year in the short run and the growth forces
unleashed through a larger capital stock and productivity that appear in the long run are
channeled through growing (private and public) consumption and investment. Consequently,
the economy receives a somewhat larger stimulus with national borrowing than with foreign
borrowing in the long run and no output reductions appear with the former way of financing the
IIJA.
Ultimately, our analysis can also be seen as the differential pattern from debt which is paid off
more evenly distributed in the short run or after a process of accumulation in the long run.
National borrowing does not necessarily imply that all the debt is paid back on a year-by-year
basis. Certainly, foreign borrowing can easily be seen as a way of financing the debt without
affecting national savings in the short run. However, we could still think of national debt
instruments whose maturity is in the long run and foreign debt channels with short term
maturity, contrary to the contrasting trends of finance we have modelled in this paper.
We believe that the main message is that a sizeable debt accumulation that has to be paid back
in a given point of time in the future will induce lack of growth in private investment and/or
consumption in the moment in which it is paid off. The latter affects different sectors of the
economy abruptly, contracting a few of them. A more regular and evenly distributed service of
the debt on a year-by-year basis will help to unleash a homogeneous growth stimulus in the long
run for all sectors of the economy. And, consequently, this facilitates forces conducive to a
widespread wage increase across all workers categories, as we are about to see. This contrast
between long run and short run indebtedness is even more relevant in the context of rising
interest rates and inflation, as well as downward revisions in GDP growth forecasts, after
Russian's invasion of Ukraine.
Manufacturing sectors benefit the most since total manufacturing exhibits higher increases than
total services of total agriculture. The impact of Construction in the short run and on Utilities in
the long run are the main drivers of this outcome. According to Lawrence (forthcoming), the
high share of investment and hence manufacturing in Asian economies is closely related to their
rapid growth (Lawrence et al., 2009). Our results seem to support the idea that an inclusive
economic growth is closely the performance of manufacturing sectors.

6. Distributional impact
According to Lawrence (2015), before the Global Financial Crisis the debate on wage inequality
revolved around the income gap between income levels (e.g., richer vs. poorer, skilled vs.
unskilled), but since then a new point of analysis has been introduced: the shares of labor and
capital in total income. In this paper we talk to the American Debate from both perspectives.

We first address the impact across occupations and wage deciles. The first columns of Table 4
show the initial shares (i.e., before the implementation of the Act) in terms of number of workers
and wages for the occupations in the model. As expected, there is an unequal relationship
between shares per number of employees (column 1) and shares per wages (column 2). Skilled

12
workers, who account for 35.3% of total workers hold 53.4% of the total wage whereas unskilled
workers represent 64.7% and account for 46.6% of the total compensation.

Job creation, and specifically jobs in the manufacturing sectors, is an important target of this
package, and it is mentioned several times by the White House (2021a). The next columns in
Table 4 show the impact of the IIJA on all occupations. The plan unleashes a general expansion
of wages and employment, with only a few exceptions remaining nearly unaffected. While
overall job creation is quite similar in the short run under foreign and national borrowing, wage
increases in the long run are considerably larger under national borrowing.

In the short run, those occupations especially related to the Construction industry experience
the largest job increases, with similar results for both scenarios (foreign borrowing and national
borrowing). Construction workers account for the 63.6%13 of the total labor in this industry, and
therefore it is the occupation that benefits the most from the increase of government spending
in Construction, with a 4.19% increase in jobs (see column 3). Craft and related trade workers
experience the second highest increases (over 0.7%) as this occupation has a high share of its
total labor force in the Construction industry. Farm, fishery & forestry workers are mostly in
Crops and Forestry and Fishery sectors. These sectors export an important share of their
production. Thus, the export decrease experienced under foreign borrowing reduces demand
for Farm, fishery & forestry workers. On the other hand, almost 90%12 of Service workers are
allocated within Health and Social services and Government and Defense services sectors. These
industries experience an output decrease under national borrowing, thus lowering demand for
service workers (0.06%). Overall, unskilled workers benefit more than skilled ones in terms of
the number of jobs created, while there is also a mild job creation for skilled ones.

In the long run employment returns to its initial level so there is no change with respect to the
initial number of jobs. Thus, adjustments in the labor market occur through wages. The IIJA
would also benefit all occupations in the long run, with larger wage increases for unskilled
occupations than for skilled ones, under foreign borrowing. Generally, workers benefit more
from national borrowing than from foreign borrowing. In general, wage increases tend to be
larger and more evenly distributed among all occupations with national borrowing than with
foreign borrowing. This is related to the production increase we have just analyzed. In the long
run, with national borrowing there was an overall expansion in production, with all sectors
expanding. With foreign borrowing the construction sector and a few services sectors would
contract, which translates in uneven increases in wages across occupations.

13
Data not included in the table.

13
 Table 4. Initial share of total wage remuneration by skill level and impact on employment and wage
by occupation and skill level under the two simulation scenarios
(% share and % change with respect to initial levels)

1. 2. Foreign borrowing National borrowing

Initial share of Initial share of Short run Long run Short run Long run
Skill total workers total wage (employment variation) (wages variation) (employment variation) (wages variation)
Occupation
level by skill level remuneration 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
by skill level Occupation Skill New share of Occupation Skill New share of Occupation Skill New share of Occupation Skill New share of
level level total workers level level total wage level level total workers level level total wage
remuneration remuneration
Skilled

Managers 0.40 2.97 0.38 4.19

35.3 53.4 0.20 35.2 2.21 53.2 0.17 35.2 3.94 53.4
Professionals 0.06 1.65 0.01 3.77
Construction 4.19 1.36 3.84 4.08
Craft & related 0.74 1.76 0.68 2.83
Sales 0.13 3.61 0.09 4.14
Unskilled

Office workers 0.23 2.35 0.19 4.00

64.7 46.6 0.52 64.8 2.99 46.8 0.47 64.8 3.94 46.6
Transport 0.27 1.92 0.27 1.82
Production 0.15 7.00 0.28 3.88
Agricultural & -0.05 4.64 0.04 3.91
Service 0.01 1.64 -0.06 6.16
Total 100.00 100.00 0.43 100.00 2.76 100.0 0.38 100.00 3.94 100.0

Source: authors’ elaboration.

Notes: results are expressed in percentage change from initial levels.

14
We now deepen into the impact that this plan would have on workers by income level deciles14.
Column 1 in Table 5 shows the percentage of total remuneration for each decile in the initial
data. The following columns show the share in the long term under each scenario. Columns 1-3
from Table 5 show the shares that each decile has in total remuneration, while columns 4-6
show the cumulative calculations that underlie the Lorenz Curve. Thus, the last row of column 1
(initial share) of Table 5 indicates that initially the top 10% of workers (i.e., the 10th decile) earn
17.8% of total compensation. In column 4 (initial share) the last cell indicates that 100% of the
workers receive 100% of the total remuneration.

Table 5. Long run impact on the shares and cumulative shares of total wage remuneration
by decile of annual average wage (%).

Cumulative share of total wage

Share total wage remuneration Decile
Decile remuneration
(cumulative
(% of
2. Long run 3. Long run % of 5. Long run 6. Long run
workers) 1. Initial 4. Initial
Foreign National workers) Foreign National
data data
Borrowing borrowing Borrowing borrowing
0-10 5.7% 5.7% 5.9% 10% 5.7% 5.7% 5.9%
10-20 6.2% 6.2% 6.3% 20% 11.9% 11.9% 12.1%
20-30 7.0% 7.1% 6.9% 30% 18.9% 19.0% 19.0%
30-40 7.3% 7.3% 7.2% 40% 26.2% 26.4% 26.3%
40-50 7.6% 7.6% 7.6% 50% 33.8% 34.0% 33.9%
50-60 8.3% 8.4% 8.2% 60% 42.1% 42.3% 42.1%
60-70 11.6% 11.5% 11.6% 70% 53.7% 53.8% 53.7%
70-80 13.5% 13.4% 13.5% 80% 67.2% 67.2% 67.2%
80-90 15.0% 14.9% 15.0% 90% 82.2% 82.1% 82.2%
90-100 17.8% 17.8% 17.8% 100% 100.0% 100.0% 100.0%
- 100.0% 100.0% 100.0%
Source: authors’ own elaboration.
Notes: deciles are defined by the % of workers. Results in green (red) indicate an increase (decrease)
from the initial share of total wage remuneration.

Under national borrowing the two poorest deciles win in terms of their slightly higher shares in
the weight in total labor remuneration (Column 3). This comes at a cost for the shares of some
intermediate workers in total labor remuneration (3rd, 4th and 6th deciles), which go down
slightly. These trends cannot be seen if we look only at the cumulative impact (i.e., Column 6),
since the good outcomes for the two poorest deciles disguise the losses for the intermediate
workers. Interestingly, the top deciles of workers remain unaffected in their shares in total
remuneration with national borrowing. Under foreign borrowing, the poorest deciles (1st and
2nd decile) maintain their shares in overall labor remuneration, while some intermediate workers
(2nd and 5th decile) increase their shares at the cost of deciles of richer workers (8th and 9th
decile). However, wages increase more under national borrowing, which makes it more
desirable for most categories of workers. Nevertheless, we should bear in mind that the overall
amount of labor remuneration has been increased, and the total remuneration distributed
across workers is larger than before, with most of them benefiting from the impact of the IIJA.

If we compare the results (long run) for the top decile (i.e., 10% of workers with the highest
wages) and the first decile (i.e., 10% with the lowest wages) before the simulations, the former
accounted for 17.8% of total wages and the latter for 5.7%. After the IIJA workers with the

14
Sorted from lowest to highest average annual wage, i.e., the 1st decile corresponds to workers with
the lowest wage.

15
highest salaries would retain their share (17.8%) in both scenarios, while the poorest would
increase their share by 0.2 percentage points, raising it up to 5.9% under national borrowing and
keeping it at 5.7% under foreign borrowing. The fact that unskilled workers wages go up by more
(2.99%) than skilled ones (2.21%) slightly reduces the gap between these extreme deciles. Since
under national borrowing wages for both skilled and unskilled experience the same increase,
their respective shares remain unaffected. However, the higher wage rise under national
borrowing should leads us to conclude that national borrowing benefits more workers than
foreign borrowing, at least in the long run15. The broad set of results provided by our general
equilibrium approach allows us to identify that the contrast between the top and the 1st decile
is not enough to evaluate the differential impact across workers16.

We finally analyze the impact that this stimulus would have on the factor incomes (wages, rents
and profits). Table 6 shows the changes in the shares of primary factors (land, labor and capital).

Table 6. Long run impact on primary factor income shares

Scenario Labour Capital Land Total

Initial data 54.38% 45.47% 0.15% 100%
Foreign borrowing (long run) 55.33% 44.52% 0.15% 100%
National borrowing (long run) 55.52% 44.33% 0.15% 100%

Source: authors’ own elaboration.

Notes: Results in green (red) indicate an increase (decrease) from the initial share.

Our results are a real example of what Lawrence (2015) highlights related to the substitution
elasticity between capital and labor: “paradoxically, with a substitution elasticity under one,
policies that increase investment and the supply of capital could achieve more equal
distributions of income.” The elasticities of substitution among primary factors that underlie our
results remain below one in all industries and scenarios. Even though our long run simulations
are based on an increase in capital stock supply and output productivity, under both scenarios
there is an increase in the share of labor income. Therefore, this plan would help to change the
U.S. declining trend of the labor-capital ratio.

7. Conclusions
We analyze the impact that an increase in public infrastructure spending has on the U.S.
economy. It is the strongest public infrastructure investment stimulus that has been approved
in the US in the last decades, going beyond the one of the Inflation Reduction Act. To the best
of our knowledge, we offer the first analysis of the Infrastructure Investment and Jobs Act (IIJA)
using a Computable General Equilibrium approach. Moreover, we present our work after the
recent approval of the plan in November 2021, while the previous analyses made on the IIJA
were carried out before it was signed into law (Bonakdarpour et al., 2021; Yaros & Zandi, 2021).
We use The Enormous Regional Model for the U.S. economy, which allows us to introduce
regional-level shocks and offer macroeconomic results, as well as industry level results.

15
Only the wages of workers in production (Plant and machine operators and assemblers) benefit
considerably more with foreign borrowing than with national borrowing, but they represent 2.5% of the
labor force.
16
For a similar result analyzing labor market results and the gender gap see Latorre (2016).

16
Furthermore, its disaggregation of employment allows us to compare results for ten different
occupations covering both skilled and unskilled workers.

We evaluate the short run effects of the increase in government demand for Construction ($110
billion after the first year), and the long run effects of the stimulus ($550 billion), once the
investments rise capital stock in the Utilities, Transport services and Information service
industries. Intuitively, it is assumed that physical infrastructure plays a key role in any economy,
but at the same time we know it requires significant upfront investment, which inquires a debt
increase. This debt can be financed through foreign debt (i.e., a trade deficit) or through reduced
national savings. We evaluate the impact under both scenarios, considering that the real impact
falls in between these two.

When the deficit is financed through foreign borrowing in the short run, the initial stimulus rises
GDP by 0.22% and employment by 0.43% but causes a domestic appreciation (GDP price index
rises by 0.65%). As a result, exports decrease by 0.72% and imports rise by 0.61%, thus,
increasing the trade deficit. As government demand for construction grows, there is a surge in
demand for construction workers (4.19%) and for those occupations with a high share of
workers in Construction, such as Craft and related trade workers (0.74%). Do note that the IIJA
will increase government spending throughout five subsequent years. So similar outcomes
would be repeated in each of those years. The trade deficit increase rises national debt, which
is paid in the long run by a lack of growth in private consumption and investment. The economy
grows through increased exports (6.61%) and reduced imports (2.45%), creating a trade surplus
to pay off its debt.

In the short run, the increase in government spending in infrastructure is not financed through
higher taxes (in neither of our simulations), but through a reduction in private consumption
(0.17%) and investment (0.35%) in the national borrowing scenario. These short run results
could be interpreted as paradoxical since the objective of increased public spending is to
improve the economy of the country and its citizens. However, we do derive that national
borrowing results in a 0.19% increase in GDP and a 0.38% rise in employment in the short run,
following the government demand stimulus. What is more, it is under this scenario in which the
most positive long-run results are obtained. GDP rises by 1.39% and wages rise by more than
3.9% compared to 1.18% and 2.76% increases in GDP and wages under foreign borrowing. We
consider it of special importance to share these results, as currently rising inflation levels and
the rise in interest rates in the near future may cause the payout of this plan to increase in a
contest of weak GDP growth after Russian’s invasion of Ukraine.

Regarding production, the short run demand increase for Construction mainly benefits this
industry, which levels up its production by more than 3.3% under both scenarios. In the long run
the economy experiences a general upward trend in production. However, we do find a
somewhat contrasting trend between national and foreign borrowing. Because in the long run
the foreign debt has to be paid back, there is a small push for production with a few sectors
reducing output. This does not happen with national borrowing. In this latter case, the impetus
of a larger capital stock is not dampened by the need to service the debt. As a result, the increase
in production occurs across the board, and this results in better results for wages across most
labor categories, with unskilled workers benefiting the most. Most wage increases for both
skilled and unskilled workers are the same (3.94%) and larger under national borrowing than
with foreign borrowing. However, inequality levels are unaffected with national borrowing
because the wages of skilled and unskilled rise by the same amount. By contrast, inequality is
slightly reduced under foreign borrowing, with the latest decile of workers according to wage

17
remuneration, slightly increasing by 0.2 percentage points their share in overall remuneration
(from 5.7% to 5.9%). The top decile retains its initial 17.8% share under foreign borrowing,
though. However, despite the slight reduction in inequality, wages go up by less under foreign
borrowing, which makes it less desirable for most workers categories. We also derive that
workers increase their share in overall GDP with capital share going slightly down in all the
scenarios we analyze.

We believe that this contrast between national borrowing and foreign borrowing can also be
interpreted as a differential impact between short term and long-term debt maturity periods.
Somehow, the main message is that the accumulation of an important debt, which has to be
paid back in a given point of time in the future tends to translate into abrupt production
adjustments and even output contraction in some sectors. By contrast, a more regular and
evenly distributed debt service tends to facilitate a strong boost for production across all sectors
and workers categories.

In sum, our results suggest that the IIJA will increase GDP, employment and wages in the US
economy, no matter how it is financed. In the short run foreign borrowing (or short run debt
maturity) results in slightly larger increases than national borrowing (or important debt
accumulation for the long run). However, in the long run the need to pay back the debt dampens
the stimulus compared to national borrowing. National borrowing (or a more evenly distributed
debt) brings about a more general push across sectors and benefits all workers, with unskilled
over benefiting the most.

18
 Appendix 1. Tables

Table A1. Model regions and regional fund allocation

Estimated fund allocation ($US

Region States
billion)
Washington, Oregon, Montana, Idaho,
NorthWest 40.58
Wyoming, Nevada, Utah
California California 57.55
SouthWest Colorado, Arizona, New Mexico 22.23
SSP Nebraska, Kansas, Oklahoma 16.34
Texas Texas 45.77
North Dakota, South Dakota, Minnesota,
Wisconsin, Iowa, Illinois, Missouri,
MISO 113.33
Arkansas, Louisiana, Mississippi, Michigan,
Indiana
Ohio, Kentucky, Pennsylvania, West
PJM Virginia, Virginia, Maryland, Delaware, 96.74
New Jersey
Tennessee, North Carolina, South Carolina,
SouthEast 81.21
Alabama, Georgia, Florida
NewYork New York 34.77

Maine, Connecticut, Massachusetts, New

NewEngland 31.74
Hampshire, Rhode Island, Vermont

AlasHawaii Alaska, Hawaii 9.75

Total 550.00

Source: authors’ elaboration based on CNBC (2021); US Census Bureau (2021)

19
 Appendix 2. Sensitivity analysis
We carry out a sensitivity analysis to check the robustness of our results following the approach
of Latorre et al. (2020b) and Latorre (2012b). To do so, we double and halve the values of several
elasticities (primary factor substitution elasticity, skill type substitution elasticity and
domestic/imported Armington substitution elasticity), as well as the value of increase in the
output productivity. Table 5 shows how the effects for GDP and labor (wages or employment)
of the IIJA are affected by each parameter. We change parameters one by one, while we keep
all the others unchanged. Additionally, we run a long run sensitivity analysis in which we isolate
the IIJA, i.e., without the long run endogenous capital adjustment (see last row). This implies we
keep fixed the capital stock in all industries, except the ones directly affected by the IIJA
(Utilities, Transport, Information). The latter do experience the capital stock increases explained
above. We run this shock because in the last decades capital stock increases have been scarce
and the long run endogenous capital adjustment may not take place in a mature economy like
the U.S.

Table A2. Sensitivity analysis: impact of the IIJA on the US

Foreign borrowing National borrowing

Short run Long run Short run Long run
GDP Employment GDP Wages GDP Employment GDP Wages
Base results 0.22 0.43 1.18 2.76 0.19 0.38 1.39 3.94
CES substitution, primary factors (σP)
σP * 2 0.25 0.48 1.28 1.63 0.23 0.46 1.61 2.42
σP * 0.5 0.19 0.36 1.02 4.10 0.15 0.3 1.17 5.08
CES substitution between skill types (σL )
σL * 2 0.22 0.43 1.19 2.81 0.19 0.38 1.38 3.90
σL * 0.5 0.22 0.43 1.18 2.69 0.19 0.38 1.40 4.02
Armington Substitution between
dom/imp (σei )
σdi * 2 0.21 0.41 1.24 2.99 0.19 0.38 1.39 3.94
σdi * 0.5 0.22 0.44 1.14 2.55 0.19 0.39 1.38 3.90
Output productivity (atot)
atot * 2 1.46 3.17 1.72 4.67
atot * 0.5 1.04 2.56 1.22 3.58
IIJA in isolation (no endogenous long run
0.74 2.01 0.74 2.19
capital adjustment)

Source: authors’ elaboration

The first row of Table A2 presents the impacts described above. Regarding short run results,
GDP increases are quite robust for any elasticity. Short run outcomes are nearly the same except
for the primary factors’ elasticity, whose impact is slightly larger. This is the parameter that has
the strongest interdependence with GDP and labor results. Long-run results show slightly higher
differences and due to the same parameter (primary factors elasticity), but small enough to
continue to demonstrate that GDP and wages will always go up vigorously. Our shock is an
expansionary one and it creates new jobs in the short run or increases wages in the long run.
When the elasticity of substitution between primary factors is larger (lower), and labor and
capital are more (less) substitutable than in the reference scenario, more jobs are created after
the increase in demand in construction. In the long run, with a larger (lower) elasticity the wages

20
increase, which are necessary to move workers across sectors, are lower (larger) after the
increase in the capital stock.

Additionally, we perform a sensitivity analysis of the productivity shocks, by doubling and

halving the ones introduced in our study. Obviously, larger productivity effects result in much
better outcomes for GDP and wages.

Finally, as for the long-term simulations in which we isolate the effect of the Biden plan, the
difference with the initial scenario lies in the mobility of the capital stock of those sectors in
which we do not exogenously increase it. That is, if the simulation is based only on the
exogenous increase in the capital stock in Utilities, Transport and Information, this increase
represents approximately a 1% increase in aggregate capital stock, while in the initial results of
our analysis the increase is above 2% in both scenarios. The reasons for carrying out this
simulation are mainly two: firstly, because in this way we are able to estimate the importance
of the sectors in which we introduce the shock (Utilities, Transport and Information) in the
results. Secondly, although intuitively/theoretically the capital stock increases in the long run in
almost all sectors, in the US the capital stock has not increased significantly in recent years 17
(Bennett et al., 2022). Therefore, it may seem optimistic to increase the capital stock by more
than 2% solely due to the IIJA.

In the case of foreign borrowing, although the aggregate increase in capital is halved in this
simulation, the increases in GDP and wages remain above 60% from initial results. With this
scenario we rule out that the greatest weight of economic growth comes from other sectors
that do not benefit from this plan. In the case of national borrowing, we reach the same
conclusion although the impact compared to the endogenous mechanism is less intense. In any
case, we still derive that this way of financing the plan benefits more the workers than in the
case of foreign borrowing.

In sum, our sensitivity analysis shows that our previous results are robust to changes in the
elasticities and other assumptions. The primary factors elasticity may change the size but not
the direction of the adjustment. In the long run, the IIJA considered in isolation brings about
important GDP and wage increases for the US economy. The impact could be more sizeable still
if the plan results in larger productivity impacts than what has been assumed based on the
literature.

17
Real basic infrastructure investment has been between $280 and $320 billion since the early 2000’s.

21
Appendix 3. The TERM Model
The U.S. version of the Enormous Regional Model (TERM-USA) is a multi-sector and multi-region
CGE model for the U.S. based on the generic TERM and developed by the Centre of Policy
Studies, from Victoria University (Australia). TERM builds on the ORANI model for the Australian
economy and provides a strategy for creating a “bottom-up” multi-regional CGE model used for
comparative-static simulations. Along with national constraints, it treats each region as an
independent economy. Thus, prices and quantities can vary from one region to another. In each
region, the households’ preferences are defined by a representative consumer, and firms
produce goods that are consumed by final users (households, government, exports) or by other
firms (as intermediate inputs). In this specific U.S. version, the 11 regions are based on an
aggregation for modelling electricity generation and distribution (electricity grid regions, see
Table A1 on Appendix 1). It distinguishes 26 sectors combining goods (17) and services (9), each
of them producing a single product or service.

As is well known, CGE models rest on the input-output structure of the economy but assume
more flexible functional forms than input-output models. Interestingly, upon national statistics,
the database of the TERM-USA model adds a regional dimension based on regional statistics.
The original database contained 2005 data, and we have updated it by imposing new targets for
regional GDP and wages (Centre of Policy Studies, 2022).

Figure A1 shows the basic structure of the TERM database. Each rectangle represents a flow
matrix whose dimension is defined by the number of set it considers, e.g., the HOUPUR (c,h,d)
is a 3 dimensional matrix as it is reflects household demand (h) in region d for commodity c. Sets
description is shown at the top right of Figure A1. Price data on each matrix can be expressed as
basic, delivered (which includes basic + trade or transport margins) or purchasers (basic +
margins + taxes) values.

The left-hand side of Figure A1 shows the production side of each region in the economy, in a
manner similar to a conventional input-output database for a region. On the top left, the USE
matrix shows industry demand for intermediates (i.e., production costs) and final user demands
(households, government, investment, exports). The sum of production costs (industry demand
for intermediates), commodity taxes (TAX matrix) and primary factor costs (labor, capital, land
and production taxes) is shown in matrix VTOT. The latest is at the same time the sum of the
value of domestic industry outputs (MAKE matrix) plus the value of inventories (STOCK matrix).
Some balancing requirements are shown in the figure, e.g., the matrix USE summed over users
(USE_U) shall be equal to DELIVRD_R, which is the sum of DELIVRD over regional sources
(Horridge, 2012). The DELIVRD matrix, in turn, gathers the prices of domestic goods and services
including trade and/or transport margins.

The matrices on the right-hand side represent the regional sourcing mechanism of the TERM
database. The TRADE matrix shows inter-regional trade of domestic or imported (s) commodity
(c) by origin (r) and destination (d). In the case of domestic source (s=domestic), “r” represents
region of origin. When the source is imported (s= imported), “r” corresponds to port of entry.
Therefore, import data shown in the IMPORT matrix is the result of adding up the imported part
of the TRADE matrix over destinations (d). The model recognizes retail trade and road transport
(i.e., margin commodities) needed to move the products between producers and final users.
TRADMAR shows the margin value needed to facilitate each flow of the TRADE matrix (Horridge,
2012), and SUPPMAR gives information on the region where each margin is produced.
TRADMAR does not include information on where the margin is produced, whereas SUPPMAR

22
is not defined for either commodities or sources (the same proportion of margin is supply
independently from the commodity). Thus, if we sum TRADMAR over commodities and sources
we obtain the same as if we sum SUPPMAR over p (regions where margin is produced).

Figure A1. The TERM flows database

Source: Horridge (2012)

As in other CGE models, the TERM model contains thousands of equations that describe the
behavior among the different agents in competitive markets (households, investors,
government, exporters and importers), using neoclassical assumptions. These equations

23
describe the equilibrium between demands and supplies and between costs and prices. Sectors
are perfectly competitive. Thus, they produce undifferentiated goods or services under constant
returns to scale technology and price at marginal costs, so that agents are price-takers.
Therefore, products/services only differ according to their region of origin in Armington fashion.
Production functions are nested CES type and assume producers act under cost-minimizing
behavior (see Figure A2).

Figure A2. TERM Production structure

Source: Horridge (2012)

Producers demand both intermediates and primary factors, following a Leontief assumption,
i.e., in proportion to industry output (Horridge, 2012). Each of the intermediate inputs’
aggregate and primary factors’ aggregate come from a CES aggregation of commodities from
different regions (intermediates’ aggregate) and a combination of land, labor and capital
(primary factors’ aggregate). Labor is at the same time a CES aggregation of 10 different
occupations (2 skilled and 8 unskilled).

24
Figure A3 represents the TERM system of demand sourcing. The rectangles on the left side of
the figure show the corresponding data matrices (shown in Figure A1) expressed as values
(uppercase) and percentage change variables of prices (prefix “p”) and quantities (prefix “x”).
Although the scheme corresponds to a specific user (Household) demand for a specific
commodity (Crops) coming from a specific region (NorthWest), this scheme applies to demands
of any commodity coming from any region.

Figure A3. TERM Sourcing mechanisms

Source: authors’ elaboration based on Horridge (2012)

As in the production structure, the TERM demand sourcing has a nested structure. At the top
nest, households combine domestic or imported varieties following a CES demand function, and
data is expressed in purchasers’ prices (PUR). In the specific example of Figure A3, the
representative household in NorthWest demands imported and domestic Crops based on a CES

25
function. Accordingly, matrix PUR_S is the result of adding up matrix PUR over sources. The rest
of matrices shown in the nested structure are not user-specific, i.e., the delivered price of a
domestic good or service (DELIVRD matrix) depends on its basic plus margin cost and it is the
same for any given user at a given region.

Domestic varieties are differentiated by region of origin under a CES assumption. Following the
example, delivered Crops from Texas (or any other region) is a Leontief composite of the basic
price of Crops plus trade costs and transport costs. The latter can be differentiated by different
circumstances, i.e., transport distance, transport modes, size, weight, etc. Thus, the margin
value depends on the combination of origin, destination, commodity and source. As mentioned
previously, the model is even more refined and considers that margins may be provided in
different regions and takes into account the transport structure of each region (demand function
follows a CES function). All users of a given good in a given region have the same sourcing mix.

All in all, the TERM model is a suitable tool for analyzing the impact of a specific public policy
measure. Based on micro and macro-economic theory, it describes the behavior and
interdependencies between the different agents of the economy. Differently from forecasting
methodologies, CGE models allow to isolate the impact of a particular policy. Moreover, the
regional component, its distinctive feature, allows shocks to be introduced at the regional level,
which cannot be done with other models.

26
 References

American Society of Civil Engineers (2021). 2021 Report Card for America’s Infrastructure.
Available at: https://infrastructurereportcard.org/wp-
content/uploads/2020/12/National_IRC_2021-report.pdf

Bennett, J., Kornfeld, R., Sichel, D. and Wasshausen, D. (2020). Measuring Infrastructure in the
Bureau of Economic Analysis National Economic Accounts. Bureau of Economic Analysis
Working Papers, WP2020-12. Available at: https://www.bea.gov/system/files/papers/BEA-
WP2020-12.pdf

Blas, E. (2010). The Dwight D. Eisenhower National System of Interstate and Defense Highways:
The Road to Success? The History Teacher, 44(1), 127–142.
http://www.jstor.org/stable/25799401

Bom, P.R.D. & Ligthart, J.E. (2014). What Have We Learned From Three Decades Of Research On
The Productivity Of Public Capital?, Journal of Economic Surveys, Wiley Blackwell, vol. 28(5),
pages 889-916, December.

Bureau of Labor Statistics (2021). May 2021 National Occupational Employment and Wage
Estimates [Data set]. Occupational Employment and Wage Statistics. U. S. Department of Labor.
https://www.bls.gov/oes/current/oes_nat.htm
Cantwell, M. (2021, October 8). Infrastructure Investment and Jobs Act will deliver on Senator
Cantwell’s Priorities. María Cantwell, United States Senator for Washington.
https://www.cantwell.senate.gov/news/press-releases/cantwell-outlines-big-wins-for-
washington-states-infrastructure-salmon-economy-

Centre of Policy Studies (2019). TPMH0182. Available at:

https://www.copsmodels.com/archivep.htm

Christensen, M., Weiers, G., y Wolski, M. The 2019 assessment of the macroeconomic effects of
the European Fund for Strategic Investments with the RHOMOLO-EIB model. Investigaciones
regionales = Journal of Regional Research, 2019, n. 45, p. 5-15. ISSN 2340-2717

Committee For A Responsible Federal Budget, U. S. (2021, July) What's in the Bipartisan
Infrastructure Investment and Jobs Act? United States. https://www.crfb.org/blogs/whats-
bipartisan-infrastructure-investment-and-jobs-act

Concrete Jungle (2022, Dec 10). The Economist, 409(8864), n/a. Retrieved from Proquest
database.

Congressional Budget Office (2016). The Macroeconomic and Budgetary Effects of Federal
Investment June 2016, https://www.cbo.gov/sites/default/files/114th-congress-2015-
2016/reports/51628-Federal_Investment-OneCol.pdf
Congressional Budget Office (2020). “The Federal Budget in 2019: An Infographic”.
Congressional Budget Office. https://www.cbo.gov/publication/56324

de Souza Ferreira Filho, J.B., Ribera, L., & Horridge, J. (2015). Deforestation Control and
Agricultural Supply in Brazil. American Journal of Agricultural Economics. 97. 589-601.
10.1093/ajae/aav004.

Euijune Kim, Geoffrey J. D. Hewings & Chowoon Hong (2004) An Application of an Integrated
Transport Network– Multiregional CGE Model: a Framework for the Economic Analysis of

27
Highway Projects, Economic Systems Research, 16:3, 235-258, DOI:
10.1080/0953531042000239356

European Commission (2014). Investment Plan for Europe. https://wayback.archive-

it.org/12090/20210128110504/https://ec.europa.eu/commission/strategy/priorities-2019-
2024/jobs-growth-and-investment/investment-plan-europe_en

Fair, Ray C. (2019). U.S. Infrastructure: 1929-2019. Cowles Foundation Discussion Paper No.
2187, July 2019, Available at SSRN: https://ssrn.com/abstract=3432670 or
http://dx.doi.org/10.2139/ssrn.3432670

Franck, T. & Rattner, N. (2021, August 31). These charts show which states will get the most
money from Biden’s infrastructure bill. CNBC.
https://www.cnbc.com/2021/08/31/infrastructure-bill-map-which-states-get-the-most-
money.html

Giesecke J., Dixon P.B. & Rimmer M.T. (2008). Regional macroeconomic outcomes under
alternative arrangements for the financing of public infrastructure. Papers in Regional Science.
87. 3-31. 10.1111/j.1435-5957.2007.00147.x.

Horridge, J. M. (2012). The TERM model and its database. In G. Wittwer (Ed.), Practical policy
analysis using TERM (1st ed., pp. 13 - 35). Springer. https://doi.org/10.1007/978-94-007-2876-
9_2

Horridge, J.M. & Wittwer, G. (2007). The economic impacts of a construction project, using
SinoTERM, a multi-regional CGE model of China, Centre of Policy Studies/IMPACT Centre
Working Papers g-164, Victoria University, Centre of Policy Studies/IMPACT Centre.

Kim, E., Kim, H. S., & Hewings, G. J. D. (2011). An Application of the Integrated Transport
Network—Multi-regional CGE Model: An Impact Analysis of Government-Financed Highway
Projects. Journal of Transport Economics and Policy, 45(2), 223–245.
http://www.jstor.org/stable/23072176

Latorre, M.C. (2012a). Impact of Foreign-Owned Companies on Host Economies: A Computable

General Equilibrium Approach. Hauppauge, NY: Nova Science Publishers, Inc, 2012

Latorre, M.C. (2012b). Industry restructuring in transition after the arrival of multinationals: a
general equilibrium analysis with firm-type costs differences, Post-Communist Economies,
Taylor & Francis Journals, vol. 24(4), pages 441-463, June.

Latorre (2016). A CGE Analysis of the Impact of Foreign Direct Investment and Tariff Reform on
Female and Male Workers in Tanzania, World Development, Volume 77, 2016, Pages 346-366,
ISSN 0305-750X, https://doi.org/10.1016/j.worlddev.2015.08.011.

Latorre, M.C., Olekseyuk, Z. Yonezawa, H. & Robinson, S. (2020a). Making sense of Brexit losses:
An in-depth review of macroeconomic studies, Economic Modelling, vol. 89, pp. 72-87.

Latorre, M.C., Olekseyuk, Z., & Yonezawa, H. (2020b). Trade and foreign direct investment‐
related impacts of Brexit. The World Economy, 43(1), 2-32.

Lawrence, R. Z. (2015). Recent Declines in Labor’s Share in US Income: A Preliminary Neoclassical

Account, Peterson Institute for International Economics Working paper WP 15-10, June,
available at: https://www.piie.com/sites/default/files/publications/wp/wp15-10.pdf

28
Lawrence, R. Z. (2022). Behind the Curve: Can Manufacturing Still Provide Inclusive Growth?, in
press.

Lawrence, R. Z., Drzeniek-Hanouz, M. & Moavenzadeh, J. (2009). The Global Enabling Trade
Report 2009. World Economic Forum. July 2009.
McBride, J. & Chatzky, A. (2019, March 8). The U.S. Trade Deficit: How Much Does It Matter?
Council on Foreign Relations. https://www.cfr.org/backgrounder/us-trade-deficit-how-much-
does-it-matter

National Highway Traffic Safety Administration, NHTSA (2017). Web Archive. Retrieved from the
Library of Congress, https://www.nhtsa.gov/sites/nhtsa.gov/files/fy2017-
nhtsa_cbj_final_02_2016.pdf

Pollitt, H., Lewney, R., & Mercure, J. F. (2019). Conceptual differences between macro-
econometric and CGE models, Cambridge Econometrics, Cambridge.
Schwab, K. (2019). The Global Competitiveness Report 2019 (p. 666). World Economic Forum.
http://www3.weforum.org/docs/WEF_TheGlobalCompetitivenessReport2019.pdf

Shirley, C. & Winston, C. (2004). Firm inventory behavior and the returns from highway
infrastructure investments. Journal of Urban Economics. 55. 398-415.
10.1016/j.jue.2003.11.001.

The White House (2021a). Fact Sheet: The Bipartisan Infrastructure Deal. Office of the Press
Secretary of the United States. 06 Nov. 2021, https://www.whitehouse.gov/briefing-
room/statements-releases/2021/11/06/fact-sheet-the-bipartisan-infrastructure-deal/

The White House (2021b). Biden Administration Releases Bipartisan Infrastructure Law
Guidebook for State, Local, Tribal and Territorial Governments. Office of the Press Secretary of
the United States. 31 Jan. 2022, https://www.whitehouse.gov/briefing-room/statements-
releases/2022/01/31/biden-administration-releases-bipartisan-infrastructure-law-guidebook-
for-state-local-tribal-and-territorial-governments/

The White House (2021c). Fact Sheet: American Rescue Plan. The White House.
https://www.whitehouse.gov/american-rescue-plan/

U.S. Census Bureau (2021). Construction Spending. Source:

https://www.census.gov/construction/c30/historical_data.html

U.S. Department of Energy, DOE (2016). FY 2017 Budget in Brief. Available at:
https://www.energy.gov/sites/default/files/2016/02/f29/FY2017BudgetinBrief_0.pdf

U.S. Department of Transportation, DOT (2022). Infrastructure Investment and Jobs Act
Authorization Table. United States. Retrieved from
https://www.transportation.gov/mission/budget/dot-infrastructure-investment-and-jobs-act-
authorization-table

U.S. Department of Transportation, DOT (2022). Bureau of Transportation Statistics,

Transportation Economic Trends, available at www.bts.gov/product/transportation-economic-
trends

U.S. Environmental Protection Agency, EPA (2016). FY 2017 Budget in Brief. Available at:
https://www.epa.gov/sites/default/files/2016-02/documents/fy17-budget-in-brief.pdf

29
U.S. Government Accountability Office, GAO (2020). Broadband: Observations on Past and
Ongoing Efforts to Expand Access and Improve Mapping Data,
https://www.gao.gov/products/gao-20-535

United States President & Council of Economic Advisers (2022). Economic report of the
President transmitted to the Congress. Washington: U.S. G.P.O. [Periodical] Retrieved from:
https://www.whitehouse.gov/wp-content/uploads/2022/04/ERP_2022_.pdf

United States Senate (2021). Senate Amendment 2137 to H.R.3684 - 117th Congress (2021-
2022). (2021, August 8). https://www.congress.gov/amendment/117th-congress/senate-
amendment/2137/all-info

Wittwer, G. and Dixon, J. (2013). Effective use of public funding in the Murray-Darling Basin: a
comparison of buybacks and infrastructure upgrades, Australian Journal of Agricultural and
Resource Economics, 57(3):399-421.

World Bank (2022). Gross fixed capital formation (annual % growth) - United States. Accessed
on May, 2022.
https://data.worldbank.org/indicator/NE.GDI.FTOT.KD.ZG?end=2020&locations=US&start=197
0&view=chart

Zandi, M. & Yaros, B. (2021). Moody’s Report: Macroeconomic Consequences of the

Infrastructure Investment and Jobs Act & Build Back Better Framework. Moody’s Analytics.
Retrieved from: https://www.moodysanalytics.com/-/media/article/2021/macroeconomic-
consequences-of-the-infrastructure-investment-and-jobs-act-and-build-back-better-
framework.pdf

Zhai, Fan. (2018). China’s Belt and Road Initiative: A Preliminary Quantitative Assessment.
Journal of Asian Economics. 55. 10.1016/j.asieco.2017.12.006.

30