Download - Wiley Online Library

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Export wikipedia , lookup

International monetary systems wikipedia , lookup

Transcript
When Are Monetary Policy Preferences Egocentric?
Evidence from American Surveys and an Experiment
David H. Bearce
Kim-Lee Tuxhorn
University of Colorado at Boulder
University of Colorado at Boulder
This article enters the international/comparative political economy debate about whether individual-level macroeconomic
policy preferences are egocentric and, if so, on what basis (factors, sectors, or firms). It argues that contextual information may
function as a precondition for the emergence of egocentric preferences. With a focus on the trade-off between using monetary
policy for a domestic or an international goal, it presents evidence from three original American surveys using informative
vignettes to show how monetary policy preferences exhibit firm-based egocentric variation: Individuals whose employer does
most of its business in overseas markets have a lesser preference for domestic monetary autonomy. It also presents evidence
from a survey experiment to show how the strength of this egocentric relationship depends on the informative power of the
vignette: A more contextually informative vignette produces a stronger relationship between overseas business activity and
a preference against domestic monetary autonomy.
R
egardless of the issue area, a complete political economy model of policymaking, especially
in a democratic context, must characterize
individual-level preferences over policy.1 If we think that
politicians take at least some cue for their policy choices
based on the preferences of potential voters, then we need
to know both what citizens want in terms of economic
policy and on what basis their policy preferences are
formed. The former is easy to ascertain, but the latter
is not. While individual-level preferences in all issue areas exhibit some variation, it has proven hard to explain
and predict this variation. The search for a framework to
understand the variation in individual-level preferences
across different macroeconomic issue areas leads to the
long-running political economy behavior debate, now in
its second wave.
The first wave began in the late 1980s as international/comparative political economists debated whether
such policy preferences were determined more by one’s
factor endowments (e.g., Rogowski 1987) or by one’s industry of employment (e.g., Frieden 1991). Both factor
endowments and industry of employment would provide
some theoretical basis for egocentric preferences, defined
as those based on personal or familial material considerations. “New” new trade theory (Melitz 2003) would also
provide a firm-based foundation for egocentric preferences, but it had not yet entered into the discussion.
Quite appropriately, this political economy behavior
debate began with theory. But it did not turn to the empirics until the 21st century, with a focus on trade policy
preferences. Using survey data from the United States,
Scheve and Slaughter (2001) found that individuals with
higher wages and more years of education are less likely to
prefer trade restrictions, consistent with a factoral model.
Using cross-national surveys, Mayda and Rodrik (2005)
found similar support for a factoral model and weaker
David H. Bearce is Professor of Political Science/International Affairs, University of Colorado at Boulder, Ketchum 134A, Campus Box 333,
Boulder, CO 80309–0333 ([email protected]). Kim-Lee Tuxhorn is Ph.D. Candidate in Political Science, University of Colorado
at Boulder, Ketchum 134A, Campus Box 333, Boulder, CO 80309–0333 ([email protected]).
Earlier versions of this article and various parts of this research project were presented at the Institute of Behavioral Science and in the
Department of Political Science at the University of Colorado at Boulder, at the 2013 International Studies Association annual meeting, at
the 2013 American Political Science Association annual meeting, and at the 2013 and 2104 meetings of the International Political Economy
Society. Our greatest thanks go to Jennifer Wolak for answering countless questions and coordinating our department’s participation in the
2012 and 2013 Cooperative Congressional Election Studies. We thank the Center to Advance Research and Teaching in the Social Sciences
(CARTSS) and Colorado European Union Center of Excellence (CEUCE) at the University of Colorado at Boulder for their financial
support, as well as Andy Baker, Stephen Chaudoin, Matthew DiGiuseppe, Jennifer Fitzgerald, and David Steinberg for helpful comments.
A codebook with a replication data set and associated do file can be found at http://dx.doi.org/10.7910/DVN/MWNBXO.
1
Here we simply generalize the opening sentence from Scheve and Slaughter (2001, 267).
American Journal of Political Science, Vol. 00, No. 0, xxxx 2015, Pp. 1–16
C 2015,
Midwest Political Science Association
DOI: 10.1111/ajps.12203
1
2
support for a sectoral model, with individuals working in
comparatively disadvantaged industries and in industries
with greater import competition less supportive of free
trade.
However, another round of survey research questioned these results. Hainmueller and Hiscox (2006) used
the same data as Scheve and Slaughter (2001) and Mayda
and Rodrik (2005) to show that their education results
stem less from human capital endowments and more
from learning pro–free trade ideas in college and graduate school. Also focused on trade policy preferences,
Mansfield and Mutz (2009) fielded two new surveys of
trade policy preferences, finding no support based on
industry exports/imports for the sectoral model and no
support based on wages/income for the factoral model.
Instead, these authors argued that trade preferences are
determined by sociotropic (national) considerations and
one’s attitudes toward various outgroups.
Together these empirical results launched the second
wave of this political economy behavior: Do economic
policy preferences have any egocentric basis at all? In
this regard, the proposition that policy preferences have
a foundation in material self-interest has long been questioned within the subfield of American political behavior
(e.g., Kiewiet 1983; Zaller 1992), although this subfield
lacked well-developed economic theories that specified
the dimensions along which egocentrism should emerge
(e.g., factors, sectors, or firms). But this same proposition
is now being questioned by international/comparative
political economists despite rich macroeconomic theories positing egocentric preferences along different dimensions.
Thus far, political economists in this second-wave
debate have tended to advance what might be read as
extreme positions: Policy preferences are either primarily
egocentric (e.g., Scheve and Slaughter 2001) or they have
no egocentric basis at all (e.g., Mansfield and Mutz 2009).
This latter position has led to the argument that economic policy preferences instead have a sociotropic basis,
defined as being determined by whether the individual
believes that the policy option is good for the nation as a
whole. But sociotropism, even if correct, does not mean
that economic policy preferences could not also have
an egocentric basis. Stated differently, sociotropism and
egocentrism are not directly competing arguments about
how individual-level preferences are formed; policy
preferences could have both a personal and a national
basis.2
2
See, for example, the results in Mansfield and Mutz (2009). To
capture sociotropism, Mansfield and Mutz use an attitudinal measure (“Perceived Effect of Trade on U.S.”), which they show to be
DAVID H. BEARCE AND KIM-LEE TUXHORN
This article thus puts aside the egocentrism versus
sociotropism debate to consider a different reason why
egocentric preferences may not always be apparent.
Since macroeconomic issues with an international
dimension may be difficult for many potential voters to
understand, they may be unable to relate personal and
familial considerations to the various policy options. But
with sufficient contextual information, individuals may
become able to connect immediate material consideration to macroeconomic policy choices, thus leading to
egocentric preferences. On this basis, egocentrism is not
“primitive” (Campbell et al. 1960, 205); instead, it must
be “informed.”
To test this argument, we need an international
macroeconomic issue with relatively low informational
content since contextual information can be added, but
not easily subtracted. Fortunately, American monetary
policy attitudes fit this consideration given that (1) the domestic monetary autonomy/exchange rate stability tradeoff is a complex issue and (2) the American public has
not engaged in a recent debate about whether monetary
policy should be used for domestic or international purposes. While the first consideration would be true in other
national contexts, the second would not in many democracies. For example, the informational content should be
higher in most Western European countries given the recent debate about whether to join the Eurozone. Hence,
our surveys and experiment use an American sample to
ascertain whether contextual information added through
survey vignettes can produce egocentric monetary policy
preferences.
Given our focus on the trade-off between directing
monetary policy toward either a domestic or an international objective, our American sample arguably represents
a “least likely” case for finding egocentric monetary policy preferences since the United States has been able to
obtain a fair amount of external currency stability without the sacrifice of domestic monetary autonomy when
other countries (e.g., China, Panama) choose to fix to the
U.S. dollar. On this basis, one could argue that if we find
informed egocentric monetary preferences in the American case, then they should be even more apparent in other
national contexts where this monetary policy trade-off is
more stark (although we do not have the space to test this
proposition within this article).
positively correlated with a pro-trade attitude. Whenever they include a parallel attitudinal measure of egocentrism (“Perceived
Effect of Trade on Self”), this egocentric measure is also positively
correlated with a pro-trade attitude. Indeed, in their National Annenberg Election Study (NAES) survey, the attitudinal egocentric
measure has more or less the same marginal effect as the attitudinal
sociotropic measure.
MONETARY POLICY PREFERENCES
As just noted, our survey instrument assesses
individual-level preferences in terms of the domestic/international monetary policy trade-off. With
contextual information provided through a vignette,
our results show that an American respondent whose
employer does more business in overseas markets has
a lesser preference for domestic monetary autonomy.
We also find in a survey experiment that the strength of
this egocentric relationship depends on the informative
power of the vignette: A more contextually informative
vignette produces a stronger relationship between the
individual’s overseas business activity and his or her
preference against domestic monetary autonomy.
These egocentric results are consistent with the predictions of “new” new trade theory as applied to the
domestic versus international monetary policy trade-off.
Thus, a second contribution of this article is to provide
some political microfoundations (Jensen, Mukherjee, and
Bernhard 2014) for the cross-national evidence showing
monetary/exchange rate policy differences. In a series of
papers, scholars (e.g., Bearce 2014; Bearce and Hallerberg
2011; Frieden 1996, 2002) have argued for such crossnational differences based on the relative political weight
of internationally versus domestically oriented producer
preferences expressed through voter and/or special interest pressure. But there has been little individual-level
evidence to support the assumption that more internationally oriented producers have a weaker preference for
domestic monetary autonomy. We fill this gap but also
show that this preference division falls more along firm
lines than along industry lines.
An Argument about Information as a
Precondition for Egocentrism
Given that sociotropic attitudes do not preclude economic
policy preferences also having an egocentric basis, we now
consider a different reason for why egocentrism may not
always be apparent. Our argument first focuses on how
contextual information may be a precondition for the
emergence of egocentric preferences. We then consider
how informed preferences might appear in terms of the
domestic versus international monetary policy trade-off:
along factoral, sectoral, and/or firm-based dimensions.
Contextual Information
In the pocketbook voting literature in American political behavior, scholars have long considered the role of
3
information. However, these various arguments have not
yet been integrated into the international/comparative
political economy behavior debate. Furthermore, even
in American political behavior, scholars have not distinguished between the types of information (contextual
versus persuasive) that are (or are not) conducive to egocentrism. Our argument takes both of these steps.
The conventional wisdom in American political behavior was that a lack of information about complicated
issues led potential voters to express “primitive” egocentric preferences (Campbell et al. 1960, 205). As Kinder
and Kiewiet (1981, 130) summarized about this conventional wisdom, “pocketbook politics requires little in the
way of political expertise. Knowing who the incumbents
are, where the polling place is located, and a few other
details are all that is needed. Given the uneven and intermittent attention the American public pays to politics,
the minimal informational demands placed upon voters
by pocketbook politics contribute materially to its attractiveness.”
Consistent with the argument to be advanced here,
this conventional wisdom about ignorance leading to
egocentrism was challenged within the American political behavior literature. As Citrin and Green (1990, 17)
observed, “most people lack the information needed to
adopt political opinions in a rationally self-interested
manner.” Previewing the type of information that might
be conducive to the emergence of egocentrism, they further noted (Citrin and Green 1990, 18, emphasis added):
“Ignorance of the costs and benefits associated with alternative policies thus mitigates the influence of self-interest
on preferences.”
Within this debate, Sears and Funk (1990, 258, emphasis added) offered a possible compromise based on
issue complexity: While information may not be a “critical precondition for self-interest effects” in certain issues,
“political issues would seem to be of only ordinary importance; perhaps information is only critical when the issues
are very complex.” Indeed, this understanding leads directly into the political economy behavior debate where
scholars seek a framework to understand individual-level
policy preference across what can be reasonably identified
as relatively complex macroeconomic topics (including international trade and money). On this basis, we expect a
lack of information to be an obstacle to the formation of
egocentric preferences in these issue areas, even if it is not
such an obstacle on simpler topics.
But what type of information should be conducive to
egocentrism? Our argument focuses on the role of “contextual” information, which must be distinguished from
“persuasive” information. The former is defined as content about the costs and benefits of a particular policy. By
4
itself, contextual information does not directly indicate
that a policy is good or bad; instead, it summarizes the relevant policy trade-offs. In the case of international trade,
the simple statement that free trade can lower consumer
prices but may also hurt certain domestic producers provides contextual information. This same statement does
not tell the reader that free trade is good or bad; it only
identifies some policy trade-offs in this issue area.
In contrast, persuasive information directly suggests
that a policy is good or bad for particular groups.3 In the
case of international trade, the alternative statement that
free trade simply lowers consumer prices could be read as
information persuasive for free trade. To the extent that
this information identifies the policy as being good for
a larger group (e.g., all consumers), it could even induce
what appears to be sociotropic (national) pro-trade preferences. Similarly, information focused on the need to
limit imports from foreign producers in order to protect
domestic workers could be read as persuasive against free
trade. To the extent that this persuasive information leads
the reader to identify with a larger ingroup of threatened
American workers, it may even induce outgroup anxiety
opposed to free trade.
Given that surveys have become the primary instrument for obtaining the policy preferences of potential
voters, it is useful to think about these two types of information in the context of a public opinion survey. While
costly in terms of both survey time and space, contextual
information can be provided through a vignette summarizing the policy trade-offs in a complicated issue area.
The goal here would be to provide enough information
about both the costs and benefits of a particular policy
so that respondents can consider the options and express
coherent preferences between/among them. To the extent
that respondents can link policy options to their material
interests, expressed preferences may be egocentric.
Alternatively, persuasive information is often delivered via a survey prime, or frame. If effective, this prime
leads respondents toward a particular policy choice, thus
homogenizing preferences and reducing the sample’s
variation for the attitudinal dependent variable. Given
this reduced variation, personal material factors entered
as independent variables should be expected to offer less
explanatory power. The key point here is that not all
information can be expected to facilitate the expression
of egocentric preferences on a complicated policy issue.
Without contextual information, expressed policy preferences on a complicated issue may simply be incoherent.
3
Persuasive information acts like a prime, or a frame, as discussed by
Hiscox (2006). In his analysis, a negative frame about international
trade is shown to produce antitrade preferences and even to obscure
the effect of egocentric factors like education.
DAVID H. BEARCE AND KIM-LEE TUXHORN
But with sufficient contextual information, egocentrism
may emerge. However, egocentrism could also be washed
out through priming, either deliberate or accidental.
Factors, Sectors, or Firms?
Our hypothesis from the previous subsection is that
citizens should be able to express coherent egocentric
monetary policy preferences given sufficient contextual
information. But what form should these egocentric preferences take? Will egocentric preferences be based on
factoral or sectoral considerations? This was the central
question in the first-wave political economy behavior debate, but it deserves some careful reconsideration with
the addition of “new” new trade theory (NNTT) to the
theoretical debate.
The Mundell-Fleming framework posits that with
international capital mobility, governments must choose
between directing monetary policy toward either a domestic objective or exchange rate stability. Domestic monetary policy autonomy refers to using interest rates (the
basic monetary policy instrument) for some internal policy goal; depending on the macroeconomic context, this
could be raising interest rates to fight inflation or lowering interest rates to spur economic growth. Alternatively,
interest rates could be directed toward exchange rate stabilization, or resisting the external pressure to move the
exchange rate in a particular direction. If domestic monetary autonomy is chosen, then there may be movement
in the exchange rate, potentially complicating international business transactions. But if the latter is chosen to
facilitate international commerce, then monetary policy
cannot be used for a certain domestic economic objective, either lowering interest rates to boost growth or raising interest rates to reduce inflation, depending on the
macroeconomic context.
The factoral model predicts that individuals who
own more of the internationally mobile factor should
more strongly favor directing monetary policy toward
an external objective (Bearce 2003, 377). Starting with
a two-factor model that includes only capital and labor,
capital is the internationally mobile factor, whereas
labor is largely confined to the domestic economy. With
their exit option, capital owners do not need to rely on
domestic monetary autonomy; they can simply move
their capital into national economies with better growth
and inflation outcomes. But without an exit option,
laborers should prefer domestic monetary autonomy
and be relatively indifferent to exchange rate instability.
Also considering a three-factor model that adds human
capital, thus distinguishing between more-skilled and
MONETARY POLICY PREFERENCES
less-skilled labor, individuals with greater education
would be expected to have lesser preferences for domestic monetary autonomy given somewhat greater
international mobility for higher-skilled labor.
This factoral model comes from the Heckscher-Ohlin
framework, which assumes that factors of production can
move easily from one sector (or industry) to another
within the national economy. But this assumption is almost certainly invalid in the short term, leading monetary
policy scholars to rely more on the Ricardo-Viner framework, with its assumption that factors cannot move easily
across sectors. Inasmuch as factors are specific to sectors
(at least in the short term), preferences should emerge
along industry dimensions. Stated differently, individuals
should all have the same policy preference within a given
industry regardless of their factor endowments.
As described by Frieden (1991), those working in sectors more confined to the national economy should have
stronger preferences for domestic monetary autonomy.
Conversely, those working in industries that do more
of their business overseas should have stronger preferences for directing monetary policy toward the exchange
rate given that currency movements may complicate their
international business transactions. This understanding
suggests that monetary policy preferences should fall
along sectoral lines, with individuals working for more
domestically oriented industries favoring greater domestic monetary autonomy. But a complication for the sectoral model is the assumption that all firms in a given
industry have the same domestic/international business
orientation. NNTT strongly challenged this assumption.
The firm-based preference model from NNTT begins
with the empirical observation that very few firms are able
to engage in international business activity; for example,
only about 4% of American firms were exporters in 2000
(Bernard et al. 2007, 105). This fact is thought to arise
because only a small number of businesses—generally
the larger and more productive firms—are able to pay
the high costs associated with entering and competing in
foreign markets. Consequently, within most industries,
only a handful of firms have an overseas business orientation, and many of these international firms produce
across different sectors and are not limited to production
defined by a single industry. Within most industries, including manufacturing sectors where production should
be highly tradable, the vast majority of firms are confined to the domestic market. NNTT thus makes a similar
prediction about preferences as does the sectoral model:
Individuals working for more domestically oriented businesses should favor greater domestic monetary autonomy. But this firm-based model expects this variation to
be based on the individual’s specific business and not to
5
be based on average industry exports/imports, which have
been the standard operational measures for testing the
sectoral model.
Having outlined the predictions from three different
political economy frameworks, we now turn to the evidence: Do individuals, with contextual information about
the domestic versus international monetary policy tradeoff, express egocentric policy preferences along factoral,
sectoral, and/or firm-based dimensions?
Survey and Experimental Evidence
on American Monetary Egocentrism
We begin to answer this question using two original
surveys with American respondents. As mentioned earlier, we anticipate that most American citizens have little
knowledge of the domestic versus international monetary
policy trade-off given that there has been no recent debate
in the United States on this subject (unlike in most European democracies, with their recent and ongoing debates
about Eurozone participation), so we have more control
over the information content within an American survey sample. We control this content in each of the two
surveys by administering an informative vignette to all
respondents.
The first survey, conducted using Mechanical Turk
(MT) in August 2012, employs a convenience sample. The
second survey employs a representative sample and was
fielded within the 2012 Cooperative Congressional Election Study (CCES) administered by YouGov Polimetrix
before and after the November 2012 national elections.
While the obvious disadvantage of the MT survey comes
from the potentially nonrepresentative sample, the advantage of using this lower-cost convenience sample is
the additional time available, which allows for a longer
and more contextually informative vignette. As one can
see in the appendix, while the two vignettes are similarly
structured, the MT vignette is 50% longer in terms of
the word count (322 versus 214) when compared to the
CCES vignette (more about the informational difference
between these two vignettes will be provided below).
Although these two vignettes differ in terms of their
length and potentially their informational strength, both
are similar in that they are designed to deliver contextual information about the policy trade-offs in this issue
area. Neither vignette includes a deliberate attempt to
prime the respondents toward a particular monetary policy preference, although there may be some accidental
framing. It bears repeating that such priming/framing,
either deliberate or accidental, should only reduce the
6
DAVID H. BEARCE AND KIM-LEE TUXHORN
sample’s expressed preference variation, thus making it
harder (not easier) to find variation along any egocentric
dimension.
The Model
While the vignettes ask the respondent to consider two
different versions of the basic domestic versus international monetary policy trade-off, our dependent variable
is the respondent’s position with regard to the following
prompt: “When the domestic economy is not growing
and the U.S. dollar’s value is weakening in the international marketplace, use the sliding scale below to indicate
which policy goal the government should prioritize using
its monetary policy.” The sliding scale has “strengthening
the U.S. dollar internationally” at one end and “promoting domestic economy growth” at the other, with higher
values (0–100) associated with a more Domestic Monetary
Preference. Since most production in the American economy does not cross national borders, it is not surprising
to observe in Table 1 that the average respondent in both
samples has a preference toward domestic monetary policy autonomy, associated with higher values of Domestic
Monetary Preference (mean = 62 in the MT sample; mean
= 60 in the CCES sample).
We focus on this question to capture the extent to
which the respondent has a preference for a domestically oriented monetary policy because our two surveys
were administered in 2012 at a time when the American
macroeconomic context included both slow growth (with
almost no threat of inflation) and a dollar that was tending to fall (not rise) relative to the currency of many major U.S. trading partners.4 Hence, the contemporaneous
domestic versus international monetary policy trade-off
was either lowering interest rates to stimulate the domestic economy or raising interest rates to stabilize the
dollar internationally. Especially given the complicated
nature of monetary policy, using a question with the correct macroeconomic context as our dependent variable is
4
Using monthly averages from the International Monetary Fund’s
International Financial Statistics, 1 U.S. dollar could be exchanged
for 1.038 Canadian dollars in June 2012, but it had fallen to parity
(1 U.S. dollar for 1 Canadian dollar) by November 2012. Likewise,
1 U.S. dollar could be exchanged for 14.39 Mexican pesos in June
2012, but only 13.04 Mexican pesos in November 2012. In terms
of major non–North American currencies, the American dollar’s
declining value can be seen over a somewhat longer time frame.
One U.S. dollar traded for 1.48 euros in May 2011, falling to 1.30
euros in November 2012. Likewise, 1 U.S. dollar bought 1.65 British
pounds in May 2011, but only 1.56 pounds by August 2012. The
U.S. currency also declined relative to the Japanese currency, with
1 U.S. dollar buying 83.46 yen in April 2011, falling to 77.93 yen in
October 2012.
important because respondents are less likely to be able
to articulate a preference with regard to a hypothetical
monetary policy trade-off, or one that does not currently
exist (e.g., the trade-off between raising interest rates to
fight inflation or lowering them to stabilize the dollar at
a more competitive level, as was the case in the early to
mid-1980s).
We use a variety of operational measures to capture egocentrism for the factoral, sectoral, and firmbased models. These multiple measures will then be
combined into a single cleaner measure for the sectoral and firm-based models. Beginning with the factoral
model, we include an ordinal measure of Income with
10 categories: (1) less than $10,000; (2) $10,000–19,999;
(3) $20,000–39,999; (4) $40,000–59,999; (5) $60,000–
79,999; (6) $80,000–99,999; (7) $100,000–119,999; (8)
120,000–149,999; (9) $150,000–199,999; and (10) more
than $200,000. To capture the variation in human capital, we also include an ordinal measure of Education with
six categories: (1) did not graduate from high school; (2)
high school graduate; (3) some college, but no degree; (4)
2-year college degree; (5) 4-year college degree; and (6)
postcollege graduate education. Our egocentric factoral
model, outlined in the previous sections, expects a negative sign on both Income and Education, as individuals
with more of the internationally mobile factors (capital
and human capital) should have a lesser Domestic Monetary Preference.
For the sectoral model, we begin with three related
industry-based measures that are matched to the respondent’s stated industry of employment/income. Using
trade data from the Bureau of Economic Analysis (2015),5
we first include a measure of Industry Exports and, second, a measure of Industry Imports, both expressed as
a share of total industry output. We also include a third
measure, a dummy variable for a Nontradable Industry, to
capture the situation when there is no reported industry
trade either in terms of exports or imports. Our egocentric sectoral framework from the previous section expects
a negative sign for Industry Exports and a positive sign for
both Industry Imports and Nontradable Industry. These
three separate measures are then combined into a single
sectoral measure labeled Net Industry Exports, which subtracts the second measure from the first (with the third
indicating the situation when Net Industry Exports = 0).
Our sectoral framework expects that Net Industry Exports
should take on a negative sign, indicating that individuals
working in sectors with greater net exports should have a
lesser Domestic Monetary Preference.
5
U.S. international trade in goods was released December 17, 2014.
U.S. international trade in services was released October 24, 2014.
7
MONETARY POLICY PREFERENCES
TABLE 1 Descriptive Statistics for the 2012 MT and CCES Samples
2012 MT Sample
2012 CCES Sample
Variable
Mean
SD
Mean
SD
Domestic Monetary Preference
Income
Education
Industry Exports
Industry Imports
Nontradable Industry
Net Industry Exports
Overseas Business = None
Overseas Business = Some
Overseas Business = Most
Overseas Business = All
Overseas Business
Union
Age
Female
White
Black
Liberal Ideology
Democrat
Republican
Unemployed
Retired
61.84
3.97
4.17
3.89
4.58
0.08
–0.69
0.58
0.34
0.07
0.01
0.08
0.25
2.81
0.46
0.66
0.05
3.46
0.52
0.25
0.03
0.04
24.04
1.98
1.33
5.22
7.71
0.27
5.20
0.49
0.47
0.25
0.12
0.28
0.43
1.22
0.50
0.47
0.22
1.00
0.50
0.43
0.18
0.19
59.58
5.27
4.02
3.76
3.60
0.33
0.16
0.71
0.22
0.04
0.02
0.07
0.28
4.68
0.50
0.80
0.08
2.91
0.41
0.28
22.12
2.11
1.47
6.62
8.32
0.47
3.56
0.45
0.42
0.21
0.15
0.25
0.45
1.24
0.50
0.40
0.27
1.16
0.49
0.45
For the firm-based model, our egocentric measures
come from a general question about the domestic versus international orientation of their particular firm: “To
what extent does your business/employer export their
production to or do business in overseas markets?” In
response to this survey item, respondents had four options: (1) none in overseas markets, (2) some in overseas
markets, (3) most in overseas markets, and (4) all in overseas markets.6 With reason to believe that the distance
between each of these categories should not be constant,
we begin with a dummy variable for each category, with
“none in overseas markets” as the omitted category. On
this basis, we expect a growing negative sign, in sequence,
for Overseas Business = Some, Overseas Business = Most,
and Overseas Business = All. Based on the expectation
that preferences for/against a domestic monetary policy should show their greatest change between Some and
6
We ask this general question because we believe that few respondents could answer or would take the time to answer a more specific question about what percent of their business activity involves
overseas markets.
Most, we also combine this information into a single firmbased measure labeled Overseas Business by combining
None and Some ( = 0) and Most and All ( = 1). Our firmbased logic expects that Overseas Business should take on a
negative sign, indicating that individuals working in firms
that do the majority of their business in external markets
should have a lesser Domestic Monetary Preference.
It is important to report that to avoid priming an
egocentric response in terms of the attitudinal dependent
variable, the queries behind these (and all of the other)
independent variables (to be described below) came after
the Domestic Monetary Preference prompt on the 2012 MT
survey. For the 2012 CCES survey, these queries came on
a different wave of the survey (separated by weeks) from
the Domestic Monetary Preference prompt. We are thus
not concerned that answering these material questions
somehow primed an egocentric response with regard to
the dependent variable.
After our egocentric independent variables, we include a limited set of control variables that are deliberately
focused on other “material” factors. Since our dependent
variable is a monetary policy “attitude,” our right-hand
8
specification does not include any other variables directly
measuring similar monetary policy attitudes following
the concern raised by Fordham and Kleinberg (2012)
about attitudes regressed on other attitudes. As a control,
we include the dummy variable Union, which measures
whether or not the respondent has ever been associated
with a labor union. We control for the respondent’s Age
using an ordinal measure with seven categories: (1) less
than 20, (2) 21–29, (3) 30–39, (4) 40–49, (5) 50–59, (6)
60–69, and (7) 70 or older. To capture any gender-based
differences in terms of monetary policy preferences, our
specification includes a dummy variable for Female. To
the extent that there could also be race-based differences,
we add dummy variables for White and Black, the two
largest racial groups in our samples.
Given the potential importance of political ideology
and partisanship, we use a five-level ordinal variable to
control for this first consideration, with higher values indicating that the respondent self-identifies as having a
more Liberal Ideology. To control for the second consideration, we also include dummy variables for respondents
identifying themselves as either Democrat or Republican
(with Independents as the omitted category). It is important to note that while these last three variables are
not, strictly speaking, material factors, they are also not
measuring directly other monetary policy attitudes.
Finally, we add dummy variables for respondents who
report being Unemployed or Retired. In our 2012 MT sample, several individuals who fit these currently nonworking categories nonetheless answered the queries behind
our Net Industry Exports and Overseas Business variables
(presumably based on their past work experience). Unemployed and Retired both drop when using our 2012
CCES sample because the question behind the latter egocentric variable was not asked of any respondent who had
self-reported as not currently working.
The descriptive statistics for these variables in both
survey samples are presented together in Table 1. In terms
of our egocentric measures, while the CCES sample is
richer, it is less international in terms of the other egocentric measures. For basic demographic characteristics,
the CCES sample is also older, more white, less liberal,
and less Democratic than the MT sample. It is also important to note that the CCES sample is not necessarily
more representative simply based on these demographic
characteristics, but it can be made so through probability
weighting (Ansolabehere and Rivers 2013), which will be
part of our statistical specification when using the CCES
sample (but not for the MT sample).
DAVID H. BEARCE AND KIM-LEE TUXHORN
Initial Survey Results
Since the dependent variable Domestic Monetary Preference is continuous between 0 and 100, our models are
estimated ordinary least squares (OLS) with robust standard errors clustered by state. Beginning with the 2012
MT sample, we estimate three versions of the model described above, presenting the results in Table 2. Our first
model uses the longer set of disaggregated egocentric variables, which are then combined for the sectoral and firmbased frameworks in the second model. Continuing with
this preferred specification, we then reduce the statistical sample to assess the robustness of our results in this
dimension.
The results are consistent across the three models.
Among the egocentric variables, neither of the factoral
measures (Income and Education) exhibits a significant
relationship with Domestic Monetary Preference. There
is some limited support for the sectoral framework, as
Industry Imports takes on a statistically significant positive
sign in Model 2.1. Likewise, Net Industry Exports takes on
a statistically significant negative sign when this combined
measure is added in Model 2.2.
But there is even stronger support for the firm-based
framework, as each of the Overseas Business category variables in Model 2.1 takes on a negative sign that gets larger
with more international activity, although only the last
two categories ( = Most and = All) are statistically significant. This result is consistent with our expectation that
the most important preference break toward a lesser Domestic Monetary Preference should occur between Some
and Most. Consequently, we proceed with our dichotomous Overseas Business variable (Most and All = 1) in
Model 2.2, and this variable also achieves statistical significance with the expected negative sign.
Using the combined sectoral and firm-based measures (Net Industry Exports and Overseas Business), we
explore the robustness of these results by restricting the
statistical sample, dropping the respondents who reported as either Unemployed or Retired. The results are
robust for each of the egocentric measures (i.e., insignificant for both Income and Education and a statistically
significant negative sign for both Net Industry Exports
and Overseas Business).
The obvious concern about the MT results in Table 2
is that they come from a nonrepresentative convenience
sample. Would one be able to observe any evidence of
material egocentrism given a more nationally representative sample of Americans? To investigate this possibility,
9
MONETARY POLICY PREFERENCES
TABLE 2 Estimates of Domestic Monetary Preference Using the 2012 MT Sample
Income
Factoral model expected sign (–)
Education
Factoral model expected sign (–)
Industry Exports
Sectoral model expected sign (–)
Industry Imports
Sectoral model expected sign (+)
Nontradable Industry
Sectoral model expected sign (+)
Net Industry Exports
Sectoral model expected sign (–)
Overseas Business = Some
Firm-based model expected sign (–)
Overseas Business = Most
Firm-based model expected sign (–)
Overseas Business = All
Firm-based model expected sign (–)
Overseas Business
Firm-based model expected sign (–)
Union
Age
Female
White
Black
Liberal Ideology
Democrat
Republican
Unemployed
Retired
Constant
Observations
R2
Model 2.1
Model 2.2
Model 2.3
–0.24
(0.48)
0.91
(0.85)
–0.04
(.20)
0.38a
(0.10)
1.51
(3.92)
–0.21
(0.50)
0.75
(0.83)
–0.19
(0.50)
0.90
(0.93)
–0.43a
(0.10)
–0.45a
(0.11)
–14.63a
(5.06)
–7.72a
(2.36)
1.23
(0.81)
2.19
(1.60)
4.83a
(2.15)
–0.32
(3.85)
–0.44
(1.25)
–0.58
(2.21)
–6.09a
(2.52)
9.94a
(3.16)
1.64
(4.49)
57.73a
(5.36)
603
0.109
–13.52a
(4.63)
–7.98a
(2.76)
0.99
(0.93)
2.43
(1.72)
4.78a
(2.02)
–0.31
(3.71)
–0.25
(1.33)
–0.26
(2.35)
–6.02a
(2.78)
–3.35
(2.52)
–14.71a
(5.64)
–28.32a
(8.38)
–7.64a
(2.60)
1.12
(0.85)
2.42
(1.65)
4.34a
(1.97)
–0.61
(3.88)
–0.18
(1.23)
–0.76
(2.21)
–5.78a
(2.44)
10.21a
(3.60)
1.39
(4.63)
56.70a
(5.61)
603
0.122
Note: Cell entries are OLS coefficients with robust standard errors clustered in the state.
a
p < .05 (one-tailed for the directional egocentric variables, two-tailed for the non-directional controls).
56.72a
(5.39)
562
0.099
10
we bought a share in the 2012 Cooperative Congressional
Election Study (CCES). While this second sample can be
treated as representative through probability weighing,
the CCES sample includes a larger proportion of older
retired Americans, and our question behind the Overseas Business variable was only asked of respondents who
had already identified themselves as currently employed.
Hence, our sample size becomes smaller when using the
2012 CCES sample.
Table 3 presents a similar sequence of Domestic Monetary Preference models using this more representative
sample. Model 3.1 uses the longer set of disaggregated
egocentric variables, which are then combined for the
sectoral and firm-based frameworks in Model 3.2. Since
our firm-based question was not asked of those who
reported as either unemployed or retired, there is no
need for an additional model to parallel the sample in
Model 2.3.
As one can quickly observe, the egocentric results
are weaker in this survey sample. For the two factoral
measures, Education becomes statistically significant, but
with the wrong sign for egocentrism. As argued by Hainmueller and Hiscox (2006), education may proxy socialization more than it captures human capital endowments.
To the extent that individuals learn the benefits of domestic monetary autonomy for the American economy, the
positive sign on Education is plausible. However, the result
is not consistent with egocentrism.
Among the sectoral variables, Industry Exports and
Industry Imports are both statistically significant in Model
3.1, but with the wrong sign for the egocentric framework
presented earlier. The same is true for Net Industry Exports in Models 3.2 and 3.3. We have no particular explanation for these unexpected results, but they appear to be
a (small) sample property and do not emerge in any of
our other survey or experimental samples.
The firm-based variables do, however, behave consistently with our egocentric expectations in the CCES
sample, although the results are weaker than observed
in the MT sample. In Model 3.1, only Overseas Business
= All takes on a statistically significant negative sign. In
Model 3.2, the dichotomous Overseas Business variable
is also statistically significant with the expected negative
sign, but its coefficient is only about half the size as that
observed in the MT sample (–7.87a versus –13.53a ).
With a particular focus on the firm-based results that
appear in both survey samples, why are these egocentric
results stronger in the MT sample than in the CCES sample? At least two possibilities come to mind. The first concerns the nonrepresentative MT sample. However, many
scholars report similar survey questionnaire results in
convenience versus representative samples (Gosling et al.
DAVID H. BEARCE AND KIM-LEE TUXHORN
TABLE 3 Estimates of Domestic Monetary
Preference Using the 2012 CCES Sample
Model
3.1
0.51
(0.68)
4.59a
(0.82)
1.38a
(0.54)
–1.11a
(0.42)
–1.10
(4.10)
Model
3.2
0.72
(0.73)
4.82a
(0.83)
Income
Factoral model expected sign (–)
Education
Factoral model expected sign (–)
Industry Exports
Sectoral model expected sign (–)
Industry Imports
Sectoral model expected sign (+)
Nontradable Industry
Sectoral model expected sign (+)
Net Industry Exports
0.93a
Sectoral model expected sign (–)
(0.44)
Overseas Business = Some
0.36
Firm-based model expected sign (–)
(3.13)
Overseas Business = Most
0.77
Firm-based model expected sign (–)
(5.51)
Overseas Business = All
–27.32a
Firm-based model expected sign (–) (10.10)
Overseas Business
–7.87a
Firm-based model expected sign (–)
(3.72)
Union
0.47
0.68
(3.37)
(3.53)
Age
–0.16
–0.62
(1.20)
(1.29)
Female
4.77
5.80
(3.18)
(3.23)
White
3.04
4.22
(6.19)
(7.27)
Black
–6.58
–5.23
(6.83)
(7.77)
Liberal Ideology
–0.53
–1.42
(1.67)
(1.68)
Democrat
2.85
0.40
(3.68)
(4.07)
Republican
–8.82 –11.11a
(4.55)
(4.23)
40.64a
Constant
36.99a
(10.30)
(8.99)
Observations
321
321
2
R
0.249
0.214
Note: Cell entries are OLS coefficients with robust standard errors
clustered in the state.
a
p < .05 (one-tailed for the directional egocentric variables, twotailed for the nondirectional controls).
2004). Furthermore, the probability weighting used in the
CCES sample to make it nationally representative serves
to make the Overseas Business results somewhat stronger,
not weaker. Thus, we are disinclined to think that the
stronger Overseas Business results in the MT sample are
11
MONETARY POLICY PREFERENCES
due, at least primarily, to peculiar sample properties, especially given relatively similar sample characteristics as
shown in Table 1 for Domestic Monetary Preference and
for the Overseas Business measures.
A second possibility concerns the strength of the informative vignette used in the two samples. As discussed
above and shown in the appendix, the MT sample received
a longer and potentially more informative vignette than
did the CCES sample due to time limitations in the latter
survey. Our hypothesis was that, given sufficient contextual information, egocentric monetary policy preferences
would emerge. Perhaps the informational content of the
CCES vignette was barely sufficient in this regard.
Vignette Survey Experiment
To investigate this possibility, we first assessed the informational content of these two vignettes in order to better
establish that the MT vignette was more informative than
the CCES vignette. From information theory, the informative content of a textual passage can be measured by
how much space is required to store the text. In this regard, the 2012 MT vignette requires 2.1 KB (50% more),
compared to only 1.4 KB for the 2012 CCES vignette.
Using Shannon’s entropy measure (the minimum number of bits to encode the information in binary form),
the MT vignette is also 50% more informative than the
CCES vignette (10,685 versus 7,120). Finally, using content analysis software from Leximancer and AlchemyAPI,
we find that the MT vignette has eight themes with 19 key
words, whereas the CCES vignette has only four themes
with 14 key words. Based on all these differences, we
argue that there are important informational differences
between our two primary vignettes.
We thus conducted a survey experiment in two waves
using Mechanical Turk (Berinsky, Huber, and Lenz 2012),
asking the same question about a Domestic Monetary
Preference.7 In this survey experiment, however, we randomly assigned each respondent either (1) no vignette at
all, (2) the less informative CCES vignette, or (3) the
more informative MT vignette. The goal here was to
determine to what extent the articulation of egocentric
firm-based monetary policy preferences depends on the
strength of the contextually informative vignette. Including both waves, our survey experiment was administered
to 1,383 American voting-age individuals, with 456 randomly receiving no vignette, 467 receiving the shorter
7
The first wave was conducted in July 2013 and the second in January 2015 (as requested by reviewers). To control for the different
macroeconomic contexts when these two waves were conducted,
our analysis includes the dummy variable Wave.
TABLE 4 Mean Values in Three Experimental
Samples
Income
Education
Net Industry Exports
Overseas Business
Union
Age
Female
White
Black
Liberal Ideology
Democrat
Republican
Unemployed
Retired
No
Vignette
CCES
Vignette
MT
Vignette
4.18
4.16
–0.50
0.04
0.25
3.00
0.46
0.77
0.07
3.49
0.52
0.20
0.03
0.04
4.28
4.18
–0.47
0.04
0.25
2.93
0.42
0.80
0.06
3.50
0.53
0.18
0.03
0.02
4.21
4.23
–0.78
0.04
0.21
2.86
0.41
0.75
0.08
3.47
0.50
0.20
0.03
0.02
CCES vignette, and 460 receiving the longer MT vignette.
Our randomization across these three treatments was effective, as demonstrated by the demographically balanced
subsamples shown in Table 4. Using a range of different
tests, there are no statistically significant differences for
any of the independent variables. We thus omit the control variables since their inclusion has no effect on the
various egocentric measures.
Using the data from this survey experiment, Table 5
presents our model of Domestic Monetary Preference for
different samples. Controlling for the vignette received,8
Model 5.1 uses the full sample, and one can observe
the expected negative coefficient for Overseas Business
(–17.56a ), whereas the other egocentric measures are either statistically insignificant (Income and Net Industry
Exports) or significant but with the wrong sign for egocentrism (Education). The next three models use only
the restricted subsamples: those receiving no vignette
8
The statistically significant negative coefficient on CCES Vignette
in Model 5.1 suggests that there may be an accidental prime within
this particular vignette that generates a somewhat weaker expressed
preference for domestic monetary autonomy. This stands as a third
possible reason for why we obtained weaker results in the 2012
CCES sample than in the 2012 MT sample; see the earlier discussion
for why priming should tend to weaken egocentric results. We
speculate that the shorter first paragraph on domestic policy goals
compared to the longer second paragraph on international policy
goals within the CCES vignette may have produced this unintended
effect (see the appendix). The comparable paragraphs are about
the same length within the Mechanical Turk vignette (although
both are longer than in the CCES vignette); correspondingly, MT
Vignette is statistically insignificant in Model 5.1.
12
DAVID H. BEARCE AND KIM-LEE TUXHORN
TABLE 5 Estimates of Domestic Monetary Preference Using Experimental Samples
First Wave
Model
Income
Education
Net Industry Exports
Overseas Business
CCES Vignette
MT Vignette
Overseas Business ×
CCES Vignette
Overseas Business ×
MT Vignette
Wave
Observations
R2
Full Sample
No Vignette
CCES Vignette
MT Vignette
5.1
–0.20
(0.30)
1.32a
(0.43)
–0.01
(0.17)
–17.56a
(3.81)
–4.13a
(1.67)
–0.82
(1.24)
5.2
–0.44
(0.49)
–0.63
(0.84)
–0.08
(0.32)
–14.02a
(6.18)
5.3
–1.03
(0.70)
1.79a
(0.86)
–0.08
(0.18)
–8.88a
(4.56)
5.4
0.89
(0.60)
2.62a
(0.81)
0.13
(0.26)
–30.10a
(6.57)
–2.28
(1.41)
1,383
0.033
–2.23
(1.94)
456
0.019
0.18
(2.61)
467
0.017
–4.99a
(2.37)
460
0.102
Full Sample
with Interactions
5.5
–0.21
(0.31)
1.27a
(0.43)
–0.002
(0.173)
–13.79a
(5.98)
–4.34a
(1.67)
–0.13
(1.32)
5.22
(7.13)
–18.01a
(8.18)
–2.31
(1.40)
1,383
0.039
Note: Cell entries are OLS coefficients with robust standard errors clustered in the state.
a
p < .05 (one-tailed for the directional egocentric variables, two-tailed for the non-directional controls).
(Model 5.2), those receiving the CCES vignette (Model
5.3), and those receiving the MT vignette (Model 5.4).
Given our relatively large sample, we are able to find
a statistically significant negative Overseas Business result in all of the treatment groups, including those that
received no vignette (Model 5.2). But while statistically
significant, the Overseas Business coefficient is relatively
weak for those who received the CCES vignette treatment (Model 5.3), consistent with what was observed in
Table 3. Indeed, there is no statistically significant difference between the Overseas Business coefficients in Models
5.2 and 5.3, suggesting that the CCES vignette was not
more informative than no vignette at all. However, the
MT vignette treatment produced a substantively strong
Overseas Business coefficient in Model 5.4 that is statistically different from the same result in both Models 5.2
and 5.3.
This latter result is confirmed using the full sample
with Overseas Business and vignette-specific interactions
terms in Model 5.5. Using this specification, the marginal
effect of Overseas Business with no vignette comes simply from the Overseas Business constitutive term, which is
statistically significant (–13.79a ). The marginal effect of
Overseas Business with the weaker CCES vignette comes
from the Overseas Business constitutive term plus the
Overseas Business × CCES Vignette interaction term. The
latter is statistically insignificant, indicating that there is
no substantive difference between providing no vignette
and the CCES vignette. Finally, the marginal effect of
Overseas Business with the stronger MT vignette comes
from the Overseas Business constitutive term plus the
Overseas Business × MT Vignette interaction term. The
latter is statistically significant, producing a negative association that is much greater than that for the other two
treatment groups (–31.80a = –13.79a + –18.01a ).
A Nationally Representative Sample
The results of our survey experiment suggest that if a
stronger vignette had been applied to the nationally representative sample, then stronger egocentric monetary
preferences would have emerged along the firm-based dimension. But we have not yet made this demonstration.
We now turn to this final empirical task. To explore the results when using a more contextually informative vignette
13
MONETARY POLICY PREFERENCES
TABLE 6 Descriptive Statistics for the 2013
CCES Sample
TABLE 7 Estimates of Domestic Monetary
Preference Using the 2013 CCES Sample
2013 CCES Sample
Variable
Mean
SD
Domestic Monetary Preference
Income
Education
Industry Exports
Industry Imports
Nontradable Industry
Net Industry Exports
Overseas Business = None
Overseas Business = Some
Overseas Business = Most
Overseas Business = All
Overseas Business
Union
Age
Female
White
Black
Liberal Ideology
Democrat
Republican
Unemployed
Retired
60.75
4.40
3.51
3.62
3.64
0.41
–0.02
0.69
0.24
0.05
0.01
0.07
0.26
4.53
0.50
0.77
0.11
2.95
0.37
0.24
0.07
0.22
21.97
2.14
1.45
6.48
8.89
0.49
4.67
0.46
0.43
0.23
0.12
0.26
0.44
1.66
0.50
0.42
0.31
1.05
0.48
0.42
0.26
0.42
in a nationally representative sample, we bought a partial share in the 2013 Cooperative Congressional Election
Study. In the 2013 CCES, we used the same vignette as
that employed in the 2012 MT sample, although we had to
drop the fourth paragraph due to time constraints.9 Following the vignette, we first asked our Domestic Monetary
Preference question and then our Overseas Business query
(in that order to avoid potential egocentric priming on
the attitudinal dependent variable). The other independent variables come from the common content questions
in the 2013 CCES. The descriptive statistics are presented
in Table 6.
In Table 7, we present the same sequence of Domestic
Monetary Preference models as shown in Table 2. What
we obtain in this nationally representative sample is a set
9
Losing this fourth paragraph does create a somewhat less informative vignette than the 2012 MT vignette, but it remains more
informative than the 2012 CCES vignette. This three-paragraph
vignette has 230 words, requiring 1.5 KB of storage with a Shannon
entropy measure of 7735. These three paragraphs also have seven
themes and 19 key words, making it almost as informative as the
four-paragraph 2012 MT vignette.
Model 7.1 Model 7.2 Model 7.3
Income
0.35
(0.54)
Factoral model
expected sign (–)
Education
1.36
Factoral model
(0.93)
expected sign (–)
Industry Exports
–0.03
Sectoral model
(0.26)
expected sign (–)
Industry Imports
–0.02
Sectoral model
(0.19)
expected sign (+)
Nontradable Industry
–1.03
Sectoral model
(2.39)
expected sign (+)
Net Industry Exports
Sectoral model
expected sign (–)
Overseas Business = Some –0.10
Firm-based model
(2.34)
expected sign (–)
Overseas Business = Most –16.96a
Firm-based model
(3.69)
expected sign (–)
Overseas Business = All
–15.93a
Firm-based model
(9.22)
expected sign (–)
Overseas Business
Firm-based model
expected sign (–)
Union
–2.38
(2.40)
Age
–0.29
(0.74)
Female
–0.73
(2.33)
White
4.84
(3.95)
Black
–2.31
(3.90)
Liberal Ideology
3.11a
(1.29)
Democrat
–1.78
(2.12)
Republican
2.02
(3.39)
0.36
(0.55)
–0.05
(0.67)
1.43
(0.89)
1.90
(1.00)
0.02
(0.20)
0.18
(0.38)
–16.66a
(3.75)
–21.54a
(2.97)
–2.30
(2.50)
–0.31
(0.72)
–0.78
(2.28)
4.96
(3.90)
–2.23
(3.80)
3.10a
(1.36)
–1.65
(2.18)
2.11
(3.35)
–2.21
(2.40)
–0.08
(0.82)
–3.68
(1.95)
5.48
(5.21)
0.75
(5.98)
2.93a
(1.43)
–1.68
(2.32)
0.93
(4.14)
(Continued)
14
DAVID H. BEARCE AND KIM-LEE TUXHORN
TABLE 7 Continued
Model 7.1 Model 7.2 Model 7.3
Unemployed
Retired
Constant
Observations
R2
3.34
(3.53)
0.83
(2.44)
45.47a
(7.37)
804
0.111
3.43
(3.49)
0.87
(2.48)
44.58a
(7.58)
804
0.111
45.79a
(9.23)
569
0.136
Note: Cell entries are OLS coefficients with robust standard errors
clustered in the state.
a
p < .05 (one-tailed for the directional egocentric variables, twotailed for the nondirectional controls).
of firm-based results that parallel those for the 2012 MT
survey. Overseas Business = Most and Overseas Business =
All are both statistically significant with the expected
negative sign in Model 7.1. In Model 7.2, the dichotomous Overseas Business indicator has a strong negative
coefficient comparable to the parallel result in Table 2
(Model 2.2). Additionally, this result gets even stronger
when we drop the unemployed and retired respondents
in Model 7.3.
Conclusion
The argument and evidence presented in this article make
two contributions to the ongoing political economy behavior debate. Regarding the first-wave debate—whether
preferences are organized along factoral, sectoral, and/or
firm-based dimensions— our survey results provide some
microfoundations in terms of monetary policy preference
models. The evidence consistently shows American preferences about whether monetary policy should be more
directed toward a domestic or an international goal to be
based around firm considerations.
The fact that there is no support for the factoral
framework in terms of monetary policy may not be surprising since this has not been the primary preference
model used by scholars in this issue area (unlike in international trade). But given the more frequent use of a
sectoral framework by monetary scholars, it is perhaps
surprising to observe that the expected preference variation between domestically versus internationally oriented
producers is more strongly firm-based than industrybased. These firm-based divisions have a foundation in
new new trade theory, which predicts that only the larger
and most productive firms would be able to pay the costs
associated with entering foreign markets. Future research
should thus directly explore this proposition with survey
questions about other firm-based dimensions (e.g., size
and productivity).
With regard to the second-wave debate—whether
macroeconomic policy preferences have any egocentric
basis—this article makes a contribution on the political behavior side by providing an intermediate position:
Economic policy preferences are not inherently egocentric or non-egocentric. Depending on the information
context, they could be either. In this regard, egocentrism
requires contextual information. But it is not too difficult to provide this contextual information even on complex macroeconomic issues like the international versus
domestic monetary policy trade-off. Thus, our experimental results suggest that while contextual information
may operate as a precondition for egocentric preferences,
advanced training in macroeconomics is not required.
A carefully written vignette could suffice. Indeed, this
understanding suggests that we might expect to observe
stronger egocentric results in surveys about international
trade and migration when contextually informative vignettes are included in the survey instruments.
Finally, the survey evidence presented in this article
sheds some light on the relatively uncontroversial nature
of American monetary policy in actual practice. In all of
our surveys, a large majority of respondents expressed
a greater preference for domestic monetary policy autonomy. This is important because American monetary
policy is set by the Federal Reserve Board (FRB), a relatively autonomous bureaucracy insulated from popular
pressure. Indeed, independent central banks, including
the FRB, have often been criticized as nondemocratic institutions (e.g., Levy 1995/1996; Pixley, Whimster, and
Wilson 2013). Yet in the post–Bretton Woods era (and
even before), the FRB has consistently set short-term interest rates in response to domestic economic conditions
(either inflation, growth, or employment) with relatively
little regard for the U.S. dollar’s stability and value in
international markets. Our descriptive data indicate that
despite its bureaucratic insulation, the FRB is nonetheless
delivering a domestically oriented monetary policy that
is broadly consistent with the preferences of the American majority. This understanding could help explain why
Americans have come to accept a nondemocratic monetary bureaucracy within an otherwise democratic political
system.
15
MONETARY POLICY PREFERENCES
Appendix: The Vignettes
2012 Mechanical Turk (MT) Vignette
The term “monetary policy” refers to the government’s
use of interest rates to address different economic problems. When the domestic economy falls into a recession,
the government could lower interest rates to stimulate
economic growth. Alternatively, when prices are rising
in the domestic economy, the government might raise
interest rates in order to fight inflation (rising prices).
The government can also use its monetary policy to
stabilize or change the U.S. dollar’s value in the international marketplace. When the dollar’s value is falling,
the government could raise interest rates in an effort to
strengthen the dollar internationally. Alternatively, when
the dollar becomes too strong, it hurts U.S. exports by
making American products seem more expensive in international markets. Thus, the government might lower
interest rates in order to make U.S. exports more internationally price-competitive.
While all of these economic problems may be important, monetary policy cannot be used to address them
all at the same time. Consider the following scenario: the
domestic economy is not growing and the dollar’s value
is falling in the international marketplace. If the government wants to stimulate economic growth, then it would
need to lower interest rates. But if it lowers interest rates,
then the dollar will only fall further in the international
marketplace. In order to strengthen the dollar, the government would need to raise interest rates, but this would
hurt U.S. economic growth.
To further illustrate this point, consider another scenario: prices are rising in the domestic economy (inflation), but the U.S. dollar is becoming too strong in the
international marketplace, hurting U.S. exports. If the
government wants to fight inflation, then it would need
to raise interest rates. But if the government wants to
make U.S. exports more price-competitive, then it would
need to lower interest rates. Once again, the government
cannot address both of these economic problems using
monetary policy because it cannot both lower and raise
interest rates at the same time.
2012 Cooperative Congressional Election
Study (CCES) Vignette
When the economy falls into a recession, the government could lower interest rates to stimulate economic growth. Alternatively, when prices are rising, the
government might raise interest rates in order to fight
inflation (rising prices).
The government could also use interest rates to stabilize or change the U.S. dollar’s value in the international
marketplace. When the dollar’s value is falling, the government could raise interest rates to strengthen the dollar
internationally. Alternatively, when the dollar becomes
overvalued, it hurts U.S. exports by making American
products seem more expensive. Thus, the government
might lower interest rates to boost U.S. exports.
Unfortunately, interest rates cannot be used to address all of these problems at the same time. Consider
the following scenario: the economy is not growing and
the U.S. dollar’s value is falling in the international marketplace. If the government wants to stimulate economic
growth, it would need to lower interest rates. But if the
government instead wants to strengthen the dollar, it
would need to raise interest rates.
Now consider another scenario: prices are rising (inflation) and the U.S. dollar is overvalued, hurting American exports. If the government wants to fight inflation, it
would need to raise interest rates. But if the government
wants to boost American exports, it would need to lower
interest rates.
References
Ansolabehere, Stephen, and Douglas Rivers. 2013. “Cooperative Survey Research.” Annual Review of Political Science 16:
307–29.
Bearce, David H. 2003. “Societal Preferences, Partisan Agents,
and Monetary Policy Outcomes.” International Organization
57(2): 373–410.
Bearce, David H. 2014. “A Political Explanation for ExchangeRate Regime Gaps.” Journal of Politics, pp. 58–72.
Bearce, David H., and Mark Hallerberg. 2011. “Democracy and
De Facto Exchange Rate Regimes.” Economics and Politics
23(2): 172–94.
Berinsky, Adam J., Gregory A. Huber, and Gabriel S. Lenz.
2012. “Evaluating Online Labor Markets for Experimental
Research: Amazon.com’s Mechanical Turk.” Political Analysis 20(3): 351–68.
Bernard, Andrew B., J. Bradford Jensen, Stephen J. Redding,
and Peter K. Schott. 2007. “Firms in International Trade.”
Journal of Economic Perspectives 21(3): 105–30.
Bureau of Economic Analysis, U.S. Department of Commerce. 2015. “BEA Interactive Tables, International Data.”
http://bea.gov/iTable/index_ita.cfm (accessed January 20,
2015).
Campbell, Angus, Philip E. Converse, Warren E. Miller, and
Donald E. Stokes. 1960. The American Voter. New York: John
Wiley and Sons.
Citrin, Jack, and Donald Philip Green. 1990. “The Self-Interest
Motive in American Public Opinion.” Research in Micropolitics 3: 1–28.
16
Fordham, Benjamin O., and Katja B. Kleinberg. 2012. “How
Can Economic Interests Influence Support for Free Trade?”
International Organization 66(2): 311–28.
Frieden, Jeffrey A. 1991. “Invested Interests: The Politics of
National Economic Policies in a World of Global Finance.”
International Organization 45(4): 425–51.
Frieden, Jeffrey A. 1996. “The Impact of Goods and Capital
Market Integration on European Monetary Politics.” Comparative Political Studies 29(2): 193–222.
Frieden, Jeffrey A. 2002. “Real Sources of European Currency
Policy: Sectoral Interests and European Monetary Integration.” International Organization 56(4): 831–60.
Gosling, Samuel D., Simine Vazire, Sanjay Srivastava, and Oliver
P. John. 2004. “Should We Trust Web-Based Studies? A Comparative Analysis of Six Preconceptions about Internet Questionnaires.” American Psychologist 59(2): 93–104.
Hainmueller, Jens, and Michael J. Hiscox. 2006. “Learning to
Love Globalization: Education and Individual Attitudes toward International Trade.” International Organization 60(2):
469–98.
Hiscox, Michael J. 2006. “Through the Glass and Darkly: Attitudes toward International Trade and the Curious Effects of Issue Framing.” International Organization 60(3):
755–80.
Jensen, Nathan M., Bumba Mukherjee, and William T. Bernhard. 2014. “Introduction: Survey and Experimental Research in International Political Economy.” International
Interactions 40(3): 287–304.
Kiewiet, D. Roderick. 1983. Macroeconomics and Micropolitics:
The Electoral Effects of Economic Issues. Chicago: University
of Chicago Press.
DAVID H. BEARCE AND KIM-LEE TUXHORN
Kinder, Donald R., and D. Roderick Kiewiet. 1981. “Sociotropic
Politics: The American Case.” British Journal of Political Science 11(2): 129–61.
Levy, David A. 1995/1996. “Does an Independent Central Bank
Violate Democracy?” Journal of Post Keynesian Economics
18(Winter): 189–210.
Mansfield, Edward D., and Diana C. Mutz. 2009. “Support
for Free Trade: Self-Interest, Sociotropic Politics, and OutGroup Anxiety.” International Organization 63(3): 425–57.
Mayda, Anna Maria, and Dani Rodrik. 2005. “Why Are Some
People (and Countries) More Protectionist Than Others?”
European Economic Review 49(6): 1393–1430.
Melitz, Marc J. 2003. “The Impact of Trade on Intra-Industry
Reallocations and Aggregate Industry Productivity.” Econometrica 71(6): 1695–1725.
Pixley, Jocelyn, Sam Whimster, and Shaun Wilson. 2013. “Central Bank Independence: A Social Economic and Democratic
Critique.” Economic and Labour Relations Review 24(1):
32–50.
Rogowski, Ronald. 1987. “Political Cleavages and Changing Exposure to Trade.” American Political Science Review 81(4):
1121–37.
Scheve, Kenneth F., and Matthew J. Slaughter. 2001. “What
Determines Individual Trade-Policy Preferences?” Journal of
International Economics 54(2): 267–92.
Sears, David O., and Carolyn L. Funk. 1990. “The Limited Effect
of Economic Self-Interest on the Political Attitudes of the
Mass Public.” Journal of Behavioral Economics 19(3): 247–
71.
Zaller, John. 1992. The Nature and Origins of Mass Opinion.
New York: Cambridge University Press.