Transcript Are tax rates lower in resource rich states?

Federal Tax Shocks and State Fiscal
Capacity: The Case of Natural
Resources
Fidel Perez-Sebastian
(University of Alicante)
Ohad Raveh
(University of Oxford)
Yaniv Reingewertz
(University of Haifa)
March 2015
Introduction
Objective
Motivation
Main results
Federal Tax Shocks and State Fiscal Capacity

Broad questions:
•
•
How do federal tax shocks affect states’ macroeconomy?
How do states respond to federal tax shocks?

These questions have been at the heart of debates in the federalism
literature.

Previous work studied the average effect:
•

Federal tax increases have a contractionary effect, mainly due to decreases in
investment (Romer and Romer 2010 and others).
Objective:
• In this work we make a first attempt at studying the heterogenous
effects of a federal tax shock across states, through which we present
a new mechanism of vertical fiscal externalities.
Motivation and Main Questions

Why should heterogeneous effects be interesting?
•
Even the most ‘equalized’ federations present vast inter-regional
differences, from income levels and prices, to levels of public good
provision (see for instance Boadway 2006, for the case of Canada).

Hence: While a federal tax shock is uniform and equal across
the nation, its effects are not.

In this paper we ask:
•
Are there systematic differences in the effects of such shocks across
states?
•
Are there any states that may actually benefit from such a tax
increase?
Fiscal Capacity


We aim to capture the differences in states’ responses to federal fiscal
shocks through differences in fiscal capacity levels.
We define fiscal capacity as the capacity to be fiscally flexible with mobile
tax bases: High fiscal capacity states have more leeway with
capital/labor tax rates.
•

Main hypothesis:
•

Such differences may lead to corresponding differences in absorption of fiscal
shocks and an unequal inter-regional competition over mobile factors.
States with relatively high fiscal capacity have better absorption of federal
tax shocks, to the extent that they may actually benefit from them.
Since fiscal capacity can result from a significant, non-mobile, source of
income, such as natural wealth:
•
We proxy fiscal capacity through natural resource abundance.
Fiscal Capacity and Resource Abundance

U.S. Case: States receive severance tax from natural resource exploitation.
•

We use the share of severance tax in total state taxes as our main proxy for fiscal
capacity.
Large cross-state variations in fiscal capacity levels; from 0 (Vermont, Rhode
Island) to more than 0.4 (Alaska).
Mechanism: Intuition

Construct a model of heterogeneous vertical tax externalities across fiscal
capacities, with two levels of government, and capital mobility.

Resource rich states exploit their fiscal advantage to present a more
competitive fiscal environment (James 2014, Raveh 2013).

As a result they are more successful in the inter-regional competition over
factors (Cai and Treisman (2005), Papyrakis and Raveh (2014), and Raveh (2013)).

These points are at the center of the theory we propose.

Model shows that with a federal tax increase:
•
•
•
Resource rich states increase their capital tax rates by less, because they have greater
tax base (in per capita terms).
Capital moves from resource poor to resource rich states.
Resource rich states benefit (may increase output), while the resource poor ones lose.
Empirics: Strategy and Results




We take the main predictions of the model to the data.
We use a maximized annual panel of U.S. states, 1963-2007.
Follow Romer and Romer (2010), and use narrative-based federal tax changes.
Look into those shocks identified as exogenous.
Main results (following a federal tax increase):
(1) Tax rates in resource rich states increase by less.
(2) Capital moves from resource poor to resource rich states (measure capital movement
using firm level data, and disaggregated tax revenues).
(3) Contraction effect is overturned in resource rich states (specifically in 8 states: Alaska,
Louisiana, Montana, New Mexico, North Dakota, Oklahoma, Texas, Wyoming).
(4) By disaggregating the federal tax shocks and state tax revenues, we see that only corporate
related shocks and revenues are responsive – motivating our focus on capital, rather than
labor.
(5) Results are robust to various resource measures, time periods, estimation techniques, and
specifications.
Literature Review: Contributions


(1) Vertical tax externalities (tax rates):
Empirics (Besley and Rosen (1998), Deveraux, Lockwood, and Redoano (2007), Esteller-More and
Sole-Olle (2001), Fredriksson and Mamum (2008), Goodspeed (2000), and Hayashi and Boadway (2001))

•
Looked into the average effect, used potentially endogenous federalrevenues/GDP measure, finding conflicting results.
•
Contribution: We study heterogeneous effects across fiscal capacities,
addressing the endogeneity through usage of narrative-based shocks.
Theory (Dahlby and Wilson (2003), Hoyt (2001), and Keen (1998)
•
Studied a common-pool problem for the average state, focusing on
elasticities of demand and the vertical channel.
•
Contribution: We add heterogenous fiscal capacities to the otherwise
standard model ; emphasize the horizontal effect of a vertical shock, and
complementarities in federal-state public good provision.
Literature Review: Contributions

(2) Tax Changes and Output (Vertical Output Multipliers)

(Alesina and Perotti (1997), Blanchard and Perotti (2002), Perotti (1999), and Romer and Romer (2010);
Ramey (2011) provides a survey)
•
Emphasized the contractionary nature of tax increases (distortionary effects,
MCPF, suppressed investment), using national-level studies.
•
Contribution: Through state-level analysis, we show such effects can in fact
be expansionary given sufficient levels of factor mobility and fiscal capacity.

(3) Federal Tax Changes and Composition of GDP

(See Gale and Orszag (2004) for a survey)
•
Showed the share of consumption and investment in GDP falls, following a
federal tax increase.
•
Contribution: We show that with sufficient fiscal capacity federal tax increases
may be positively associated with state-level investment.
The Model
Framework

Zodrow and Mieszkowski (1986) capital tax competition model, with a
federal government and heterogeneous states with varying fiscal
capacities.

Consider a federation (keeping the U.S. case in mind) of N states,
competing for national capital stock using fiscal means.


Nash game: characterize states’ best response to federal tax changes,
taking other states’ behavior into account.
Study the consequent capital reallocation across the nation.
Assumptions

Closed economy (simplifying assumption):
•

Focus on capital:
•
•

More relevant to the shock studied (increases in federal tax rates); what we need is that factors are more mobile
within the federation than inflows from abroad (applicable via Kalemi-Ozcan et al. (2010)).
Labor less responsive to federal tax shocks; i.e. labor prefers densely populated, urbanized areas (Michaels et al.
QJE 2012).
Substantiate this in empirical part (providing evidence this is story is about capital specifically).
Federal and state public goods are complements (not substitutes):
•
Federal vs. State government spending once pensions and debt interest payments are subtracted
from it, 1960-2010:
•
Evolution suggests complementarity: correlation of 0.98.
Setup



Each state ‘i’ is populated by a fixed mass of consumers of size 1.
Each individual owns a fixed amount of capital ‘k’ that can be supplied to the
production activity in any of the states.
State ‘i’ consumers obtain an interest rate ‘r’, and these proceeds net of taxes are
allocated to the purchase of a private good ‘c’.

There are federal and state governments; both tax capital. The former (latter) sets
tax rate ( ) to finance ( ).

State governments collect severance tax on natural resources of


.
Profit-maximizing firms in state ‘i’ are owned by state residents, and produce
output (y) using capital (k) according to the following production function:
Profits (π) from firms’ activity are distributed among their owners.
Characterizing Equilibrium

Capital flows between regions until:

Market clearing conditions:
Region’s Problem

Nash game: State ‘i’ chooses to maximize the representative consumer’s utility
taken the behavior of other governments into account.
is the elasticity parameter, and lies in

•
•


.
When
: Perfect compliments
When 1: Perfect substitutes
is a government inefficiency parameter, and lies in (0,1).
Federal public good provision:
Region’s Problem

FOC:

What do we learn from this?

1)
•

if
(that is if ‘G’ and ‘g’ are not perfect substitutes)
In this case the resource rich state/s will have lower tax rates, and greater capital per
capita.
2) can go up or down with
between the public goods.
•

decreases with
If
, depending on the degree of complementarity
is sufficiently negative, the two tax rates go in the same direction.
Some more algebra leads to:
•
As
goes to minus infinity the LHS converges to 1.
How Do State Tax Rates Respond to a Federal
Tax Shock?

Using the implicit function theorem we get:

Following the previous result, as the two goods become more complements:

First, the partial derivative is positive.
Second, it will be lower for states with greater capital per capita (resource rich
states)


Resource rich states win the competition over capital due to their resources, and
hence can increase tax rates by less (following a federal tax increase), given their
bigger tax base, in per capita terms.
•
This will then lead to capital flows to the resource rich states, and corresponding differences in
output.
Conclusions from Model


If G and g are sufficiently complementary, and capital per capita is higher
in resource rich states, then following a federal tax increase we would
expect to see:
•
1) State tax rates increase more in resource poor states.
•
2) Following (1), capital flows from resource poor to resource rich states.
•
3) Following (2), state output increases in resource rich areas (if positive effect
of capital inflows is larger than the negative government inefficiency effect).
However, prior to (1)-(3), we would also like to realize whether:
•
•

(i) Tax rates are in general lower in resource rich states.
(ii) Capital per capita is higher in resource rich states.
We look into (i)-(ii), and (1)-(3) in the empirical part.
Empirical Part
Empirical Part: Objectives

In this section we test the model’s assumptions and predictions,
through the case of the US.

Specifically, we ask:
•
•
‘Preliminary’ questions:
–
Are tax rates lower in resource rich states?
–
Is capital per capita higher in resource rich states?
Main analysis:
–
Do resource rich states make a relatively smaller increase in their tax rates,
following a federal tax increase?
–
Does capital flow to resource rich states, following a federal tax increase?
–
Does output increase in resource rich states following a federal tax increase?
Data

We use an annual panel, over the period of 1963-2007, for the 50 U.S. states.

Federal tax shocks: narrative-based measures of tax shocks, derived from Romer
and Romer (AER 2010).
•
Using narrative-sources decompose changes in federal tax rates to endogenous and
exogenous.
•
Studied all significant tax acts from 1945 to 2007 (around 50), and classified their
motivation and their approximate effect on revenues.
•
Sources: “Economic reports of the President” ; “Report of the Secretary of the Treasury”
; “Budget of the US Government” ; Presidential speeches ; and more.
•
Four categories of motivation for tax changes: spending-driven, countercyclical action,
inherited-deficit-driven, and long-run-growth-driven. Consider the last two as
exogenous.
•
We use the exogenous shocks in our analysis; divide them to corporate/non-corporate
related (based on nature of act).
Data
Exogenous federal tax shocks, 1963-2007:
-150
-100
-50
0
50
100

1960
1970
1980
1990
2000
year

Variation across years (increases, decreases, and no change).
2010
Data

State characteristics:
•

States’ income data:
•

Number of firms and establishments (1977-2007). Source: Census Bureau.
Inter-regional migration data:
•

Use data on state tax revenues (including: severance, corporate, non-corporate, and total tax
income), and transfer payments. Source: Census Bureau.
Firm level data:
•

Population, GSP, etc. ; Source: Bureau of Economic Analysis.
Inter-regional labor flows (2000-2007). Source: Census Bureau
Additional resource abundance measures:
•
Mining output and oil prices. Source: Bureau of Economic Analysis.
Preliminary Questions

Are tax rates lower in resource rich states?

Previous work:
•
•
Raveh (2013): Show in a cross-section of US states (1977-2008) that: 1) Tax rates are lower in
resource rich states ; 2) The better fiscal environment attracts capital [60% of resource-induced
capital inflows are due to the business environment].
James (2014): Show in a panel of US states (1958-2008) that tax rates are lower in resource rich
states.

A cross-section of the Corporate Business Tax Climate Index (2006-2011 average):

We present similar results in our analysis.
Preliminary Questions

Is the capital tax base (on per capita terms) larger in resource rich states?

Do not observe capital stocks directly; can make inferences on capital tax base
through corporate tax revenues.

Estimate:

Corporate tax (rates) revenues per capita are (lower) higher in resource rich states =>
Indicates capital tax base is higher.
Empirical Model

We estimate models of the following type, for state ‘i’ at year ‘t’, 1963-2007:

LHS: annual growth rate in an outcome variable (average tax rates, tax revenues
per capita, inter-state migration, firm movement, and output).

RHS:
•
•
•
•
•
•
•

Lagged dependent variable, in levels.
Federal tax shock (total, corporate-related, non-corporate related), normalized by GDP.
Fiscal capacity measure: share of severance tax in total tax revenues.
Interaction between federal tax shocks and fiscal capacity.
Vector of controls ‘X’ that includes: GSP per capita, average GSP per capita of all other states
(horizontal channel), transfers from central government, population, and a ‘deductibility’ dummy.
Time trend.
State fixed effects.
Focus on
•
.
Besides next exercise, focusing on
.
Heterogeneous Vertical Tax Externalities


Are tax rates in resource rich states less responsive to federal tax shocks?
Examine average tax rates (Tax Revenues/GSP)
•

Fiscal capacity threshold (mean): severance-tax/total-tax > 0.03 ; results hold for using
median (0.005) as threshold.
Results follow model’s prediction: Tax rates are less responsive in resource rich states.
Factor Movement: Tax Revenues

Does capital flow to resource rich states following a federal tax hike?
•

Magnitude: Having 20% fiscal capacity, a 1% increase in corporate-federal-revenues/GDP increases
growth rate of corporate tax revenues per capita by 0.05%.
•

As an initial step examine corporate/non-corporate tax revenues (per capita).
Relevant for: Alaska, Louisiana, Montana, New Mexico, North Dakota, Oklahoma, Texas, and Wyoming.
Corporate tax revenues increase in resource rich states – despite having lower tax rates: tax
base may increase.
Factor Movement: Firms and Labor


Does capital flow to resource rich states following a federal tax hike?
Look into inter-state labor flows and firm movement.

Magnitude: Having 20% fiscal capacity, a 1% increase in federal-revenues/GDP increases growth rate of
number of firms (5+ employees) by approximately 0.1%.

Results suggest model’s predictions are valid: Firms move to resource rich states.
Heterogeneous Output Effects

Can output increase in resource rich states following a federal tax hike?

Magnitude: Having 20% fiscal capacity, a 1% increase in federal-revenues/GDP increases growth rate of
output by approximately 0.06%.

Results follow model’s prediction: Negative effect mitigated in high fiscal capacity states, to the
extent of being reversed.
Robustness

Results are robust to:
•
Different measures of fiscal capacity:
–
–
Mining share in GSP
Oil price: Mining share in GSP in initial year multiplied by price of oil at time ‘t’.
•
Excluding all high fiscal capacity states (specifically, those with share of
severance tax in total tax revenues greater than 10%).
•
Different estimation techniques:
–
–
•
Level regressions instead of growth-based ones.
Arellano-Bond.
Different specifications:
–
–
Inclusion of year fixed effects.
Dropping controls, excluding fixed effects, etc.
Summary of Empirical Results

Start with:
•
Tax rates are lower and capital tax base (per capita) is higher in resource rich states.

Only corporate-related federal tax shocks matter for heterogeneous effects.

Following a corporate-related federal tax hike, we observe:
•
State tax rates increase only in resource poor states.
•
Corporate tax revenues (per capita) increase in resource rich states, suggesting an
increase in capital tax base.
•
Number of firms increases (5+ employees) in resource rich states.
•
No effects on labor movement.
•
Overall the negative output effect is mitigated in resource rich states, to the extent that it
reverses and becomes expansionary.
Extensions

Does this mechanism have anything to say about the effect on the national
economy?

If capital reallocation across the nation is inefficient, it does.

Resource abundant states are less agglomerated (Perez-Sebastian and Raveh
2014):

Mechanism suggests capital flows from agglomerated to non-agglomerated areas,
thus losing agglomeration externality – dropping national output.
Conclusion

How do states respond to a federal fiscal shock?
•
Uniform shock across the economy, with differential effects.

Studied heterogeneous effects across fiscal capacities (proxied by
resource abundance).

High fiscal capacity states have a better absorption capacity for
federal fiscal shocks.

Given a corporate-related federal tax hike these states can benefit
from capital inflows that translate to increases in output, on the
account of the other states.

These results highlight a new aspect of the equalization-related
challenge dealt by federal governments.