View original document

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

Economic Review
Federal Reserve Bank of Dallas
May 1986


Fiscal Pressure and Central Bank
Policy Objectives
Richard C. K. Burdekin

Although the Federal Reserve is formally independent of
government, this article suggests that fiscal pressure from
the administration nevertheless exerts a highly significant
effect on u.s. monetary policy. In the model of central
bank behavior, fiscal pressure is proxied by the cyclically
adjusted budget deficit. The response to this deficit then
interacts with the emphasis placed on the competing goals
of monetary policy. One of the findings is that the pricestability objective tends to be compromised at higher levels
of the deficit. The indicated pattern of Federal Reserve
behavior is stable over the full 1961-83 sample period.

Energy's Contribution to the Growth
of Employment in Texas, 1972-1982
John K. Hill

An input-output analysis of the Texas economy confirms
the established opinion that energy-producing industries
were important to the state's growth during the 1972-82
period. Including multiplier effects, the employment gains
in oil and gas extraction and oil field machinery manufacturing can account for almost one-half of the increase in
Texas employment. Still, the analysis leaves unexplained a
significant portion of the state's manufacturing growth.
This suggests that other locational attributes were also
important, providing some basis for optimism regarding
the long-term prospects for growth in the state.

This publication was digitized and made available by the Federal Reserve Bank of Dallas' Historical Library (

Economic Review
Federal Reserve Bank of Dallas
May 1986
Robert H. Boykin
First Vice President
William H. Wallace
Senior Vice President and Director of Research
Harvey Rosenblum
Vice President and Associate Director of Research
James E. Pearce
Assistant Vice President and Senior Economist
Leroy O . Laney
Eugenie D. Short

W Michael Cox
Gerald P O'Driscoll, Jr
Robert T Clair
John K. Hill
Richard C K Burdekin
Steven L Green
Stephen P. A Brown
William C. Gruben
Ronald H Schmidt
Hilary H. Smith
Roger H Dunstan
William T Long III

Virginia M . Rogers
Elizabeth R. Turpin
Graphics and Typesetting
Graphic Arts Department

The Economic Review is published by the Federal
Reserve Bank of Dallas and will be issued six times
in 1986 (january, March, May, July, September, and
November) The views expressed are those of the
authors and do not necessarily reflect the positions
of the Federal Reserve Bank of Dallas or the Federal
Reserve System.
Subscriptions are available free of charge Please
send requests for single- and multiple-copy subscriptions, back issues, and address changes to the Public
Affairs Department, Federal Reserve Bank of Dallas,
Station K, Dallas, Texas 75222 (214/651-6289)
Articles may be reprinted on the condition that the
source is credited and that the Research Department
is provided with a copy of the publication containing
the reprinted material

Fiscal Pressure and Central Bank
Policy Objectives
Richard C. K. Burdekin
Visiting Scholar

Federal Reserve Bank of Dallas

The Federal Reserve has...been able to fashion monetary
policy in an impartial and objective manner-free from any
sort of partisan or parochial influence. While the long history of the Federal Reserve is not faultless, its policies have
consistently been managed by conscientious individuals
seeking the nation's permanent welfare- rather than today's fleeting benefit.
-Arthur F. Burns, Reflections of
an Economic Policy Maker
It seems to me highly dubious that the United States, or for
that matter any other country, has in practice ever had an
independent central bank in this fullest sense of the term.
Even when central banks have supposedly been fully independent, they have exercised their independence only so
long as there has been no real conflict between them and
the rest of the government.

-Milton Friedman, "Should There Be an
Independent Monetary Authority?"
Statutory independence from government is enjoyed in
some measure by each of the central banks entrusted with
the execution of monetary policy in industrialized economies. An important feature of the institutional setting concerns the longer planning horizon that may pertain to the
central bank by virtue of both the freedom from reelection
considerations and the long terms of office typically granted
to the governors of the bank. This longer horizon implies
an incentive to attach relatively more weight to the eventual inflationary consequences of expansionary policy than
Economic Review - May 1986

is the case with the elected government, which may instead
be more concerned with the possible short-run gains in
employment associated with an expansionary policy. As
discussed in the accompanying box, the fact that the benefits of anti-inflationary policy tend to be delayed, while the
costs are more or less immediate, may suffice to make such
a policy more attractive to the central bank than to the
This perspective on the relative policy goals of central
bank and government is used here as the keystone for an
analysis of u.s. monetary policy over the post-1960 period.
Monetary policy in this case is taken to be affected not just
by the unconstrained policy objectives of the central bank
but also by the influence of fiscal pressure stemming from
the government. Fiscal pressure is the means by which the
government may exert an indirect influence on monetary
policy, with the objective of the pressure being to move
monetary policy closer to that which would accord with the
government's own preferences. Central bank response to
this pressure is motivated by recognition of the government's ability to undertake legislative reform that would re1
move the formal independence of the central bank.
The suggested framework is one in which independent
central bank stabilization objectives and response to fiscal
pressure are joint determinants of monetary policy. In the
empirical work, fiscal pressure is proxied by a measure of the
full-employment budget deficit. Strong support is obtained

Inflation as a Central Bank Policy Criterion
The initial effects of anti-inflationary policy generally include upward pressure on interest rates and an increase in
unemployment. It is frequently asserted, however, that in
the long run, these adverse effects will be removed and the
effect of a tighter monetary policy will be confined to a
permanent reduction in the inflation rate. The policymaker's time horizon is likely to be very important in determining whether anti-inflationary policy is evaluated in
terms of its long-run benefits or short-run costs. In turn,
central bank freedom from reelection considerations, together with the long terms typically granted the governors
of the bank, may lead to a greater willingness to undertake
anti-inflationary policy than is the case with the government.
Moreover, if the central bank's reputation for promoting
price stability exerts an important influence on the longterm inflation rate, then concern for this reputation itself
would be a factor discouraging the central bank from resorting to overexpansionary policy. In a model developed
by Robert Barra and David Gordon, it is shown that the
policymaker's discount rate is crucial in determining
whether the policymaker forgoes the potential current
benefits of expansionary policy in favor of securing the gain
of low average inflation over the longer haul. Expansionary
policy here may temporarily lower the unemployment rate
but ultimately results only in an increase in the inflation
rate, with unemployment reverting to a long-run equilibrium level that is taken to be independent of monetary
policy. In relating the analysis to the respective time horizons of the central bank and government, it is most relevant that, other things equal, the lower the discount rate,
the closer is the outcome to the model's optimal solution,
in which the policymaker follows a zero money growth
rule. 1
The potential importance of an independent monetary
policymaker has been examined by King Banaian, Leroy
Laney, and Thomas Willett, who obtain cross-sectional evidence suggesting that central bank autonomy exerts an
independent influence on the rate of inflation. The indicated significance of dummy variables for the highly autonomous central banks of Switzerland, the United States,
and West Germany also supports the Michael Parkin-Robin
Bade finding that these banks, which are independent of
central governments both in policymaking and in the appointment of directors, have delivered a lower rate of inflation than other central bank types. 3
However, acceleration of inflation in the post-World War
II period has taken place even in countries having formally
independent central banks. This in itself may be indicative
of the pressures for monetary expansion to which central
banks are subject. The particular pressure stemming from


the national government and fiscal policymaker has been
addressed by Thomas Sargent and Neil Wallace, in whose
model the time paths of both government spending (apart
from interest payments on government debt) and tax revenues are taken as fixed 4 Here, the government is unwilling (or unable) to keep the debt-to-income ratio from rising
over time . Financing requirements dictate that a current
Increase in the stock of government debt must be followed
by more rapid money growth either now or at some point
in the future. The implication of this setting is that although
fighting current inflation with tight monetary policy may
work temporarily, such a strategy must eventually give way
to higher inflation and delayed monetization of the deficit.
The above discussion relates to a scenario in which fiscal
policy is taken as given, and there is a one-way direction
of causality running from the fiscal authority to the monetary authority. The pressure arising from the prior setting
of fiscal policy, then, may limit the central bank's ability to
pursue its price-stability objective This approach can be
generalized to incorporate the full range of stabilization
goals, and In the main text, allowance is made for interaction between the expansionary impact of the budget
deficit and the more conservative trade-off weights to
which the central bank may adhere.

Robert J Barro and David B Gordon, "Rules, Discretion and
Reputation in a Model of Monetary Policy," Journal of Monetary
Economics 12 (July 1983): 101-21


King Banaran, Leroy O . Laney, and Thomas D . Willett, "Central
Bank Independence: An Internatronal Comparison," Economic
Review. Federal Reserve Bank of Dallas, March 1983, 1-13, and
"A Comparative Study of Central Banking Arrangements and
Inflation in Industrial Countries," in Political Business Cycles: The
Economics and Politics of Stagflation, ed Thomas D . Willett (San
Francisco: Pacific Institute for Public Policy Research, forthcoming summer 1986)


Michael Parkin and Robin Bade, "Central-Bank Laws and Monetary Policies: A Preliminary Investigation," in The Australian
Monetary System in the 19705, ed Michael G. Porter (Clayton,
Australia: Monash University, 1978), 24·39, and especially "Central Bank Laws and Monetary Policy" (Revised paper, Department of Economrcs, L.;niversity of Western Ontario, London,
Ont, and Hoover Institution, Stanford, CaliL, September 1979)


Thomas J. Sargent and Neil Wallace, "Some Unpleasant
Monetarist Arithmetic," Federal Reserve Bank of Minneapolis
Quarterly Review, Fall 1981, 1-17. Further discussion is provided
by two articles in the Spring 1984 Federal Reserve Bank of
Minneapolis Quarterly Review: Michael R. Darby, "Some Pleasant
Monetarist Arithmetic," 15-20, and Preston J. MIller and Thomas
J Sargent, "A Reply to Darby," 21-26

Federal Reserve Bank of Dallas

for the hypothesized effect of fiscal pressure on the trade-off
weights applied to the competing goals of monetary policy.
The empirical performance of the model is particularly notable in indicating a stable pattern of Federal Reserve policy
over the full 1961-83 period, in contrast to the wide-ranging
instability noted in many other studies. 2
Analysis of the institutional setting

In focusing on the difference between the time horizons
relevant to central bank and government, the longer-term
outlook arising from the structure of the central bank can
be related to the "planner" function that is considered in the
context of a model developed by Richard Thaler and H. M.
Shefrin. 3 For the individual economic agent, the characterization of a "planner" concerned with lifetime utility is contrasted with that of a "doer" existing only for one period and
being completely myopic. Here, the institution of central
banking can be interpreted as an instrument of control, with
the length and staggered nature of the bank officials' terms
of office seen as being associated with incentives to consider future implications of current policy.
Meanwhile, the government remains subject to shortterm incentives related to the transitory benefits from
expansionary policy and to the attractiveness of inflationary
finance as a means of reducing the real value of government
debt. The government would, therefore, be characterized
in this context as the doer, constrained by the institutional
structure from direct intervention in the process of monetary policy.
Although the set of rules defining central bank independence represents a less than definitive constraint-given the
government's ultimate scope for instituting legal reform and
removal of the formal independence-the costs involved in
removing independence imply that only under conditions
of persistent conflict would such steps be in the government's own best interests. At the same time, to the extent
that the central bank recognizes the existence of conditions
of conflict sufficient to induce retributive action on the part
of the government, it would be rational for the bank to limit
the exercise of its discretionary powers so as not to threaten
its autonomous status.
This paper does not directly address the optimality of the
limited form of "independence" that may be enjoyed by the
Federal Reserve System. Kenneth Rogoff, for example,
shows that in the presence of a labor market distortion, it is
optimal to have a central bank that places a greater weight
on inflation than does society as a whole. 4 Under this approach, fiscal pressure that reduces the central bank's
commitment to fighting inflation would have the disadvanEconomic Review-May 1986

tage of increasing the equilibrium rate of wage inflationwith wage setters seeking pay increases that will compensate them for higher expected rates of inflation.
On the other hand, a situation in which the Federal
Reserve has less than complete autonomy may alleviate
certain technical defects associated with a genuinely independent central bank. Milton Friedman describes such
technical defects as being dispersal of responsibility, "an extraordinary dependence on personalities," and an undue
emphasis given to the point of view of bankers. s These
considerations at least suggest that, in terms of the optimal
institutional setting, the question of too much independence may be as relevant as that of excessive subordination
to the government.
Development of the empirical procedure

The central bank and government are each assumed to be
concerned with stabilization of a common set of economic
variables. The objective is to ensure that such key economic variables as inflation and unemployment stay as
close as possible to given target levels. In view of trade-offs
between the different policy goals, however, the actual policy preferences remain a function of the relative weights
adhered to by each body in assessing the importance of,
say, employment stability versus price stability. A particularly relevant issue is that the inflationary effects of an
expansionary policy are generally delayed, whereas any
positive impact on employment tends to be much more
immediate. Given this setting, a difference in the time horizons appears sufficient to imply that the government will
indeed attach more weight to unemployment than the
central bank and less weight to inflation.
The ability of the government to apply pressure to the
central bank indicates that monetary policy will be a function of both sets of trade-off weights. (Details of this approach are discussed in the Appendix.) That is, the stress
that the central bank would otherwise place on price stability is modified by government pressure arising whenever
monetary policy departs from the strategy implied by the
more short-term objectives of the fiscal authorities. Indeed,
as the central bank responds to this pressure, it could in effect be viewed as compromising its unconstrained trade-off
weights so as to conform more with those of the government and, in this way, alleviate the source of the pressure.
With the deficit serving as the proxy for fiscal pressure,
government preferences are then infused into the monetary
policy process to an extent that varies positively with observed data on government fiscal policy, the precise measure being the cyclically adjusted deficit divided by trend

GNP (OEF).6 Monetary policy is itself represented by the rate
of growth of the monetary base (OMB). The goals of price,
interest rate, and employment stability are represented by
series on the rate of growth of the GNP price deflator (OP),
the three-month u.s. Treasury bill rate (TB), and the unemployment rate (UN). Furthermore, the composition of federal spending, as reflected in the rate of growth of real
government purchases (OG), is introduced into the model. 7
The response to these variables is then taken to interact
with the deficit, the result being that each economic variable is placed alongside a corresponding interaction term in
the equation to be estimated. 8 Including lagged values of
the monetary base and the deficit taken separately, this
equation has the form set out below:

OMBt = Po




PgOMBt _ g +




g= 1



0i1 OCt -


Estimation results for the United States

+ 0i2(OEF· OC)t -


i= 1




Kj1 OPt -



Kj2(OEF· OP)t -





ek1 TBt - k + edOEF • TB)t - k




'!/1 UNt -I

+ '!/2(OEF· UN)t -



where the lag structure of the interaction terms is stipulated
as being synonymous with that of the basic economic variables. Greek letters in the equation represent parameters
that are to be estimated; u t is an error term.
The hypothesis that the government places less weight on
price stability than the central bank and more weight on
unemployment implies that the interaction effects for both
inflation and unemployment will be positive. That is, in the
case of the policy response to higher inflation, fiscal pressure has an expansionary effect by offsetting the contraction in monetary base growth otherwise following from
the central bank's desire to bring down inflation. In the case
of the policy response to increased unemployment, the
positive countercyclical response is exacerbated as fiscal
pressure induces the central bank to place more weight on
this stabilization objective. 9

In general, the theory implies that fiscal pressure leads to
the countercyclical response being exacerbated for the
variables to which the administration attaches relatively
greater weight. At the same time, the countercyclical response should be damped with respect to policy objectives
that are of relatively greater concern to the central bank.
Therefore, it is also true that if the central bank cares relatively more about rising interest rates than does the administration, the theory implies that the interaction effect
should be negative with respect to the Treasury bill rate.
This proposition has some justification because the central
bank has a banking "constituency" whose interests must
necessarily be closely tied to the state of financial markets. 10 At the same time, the hypothesized response to
higher interest rates remains less clearly defined than is the
case for the inflation and unemployment variables. No
strong a priori arguments can be advanced in respect of
government purchases.

In applying the model to the u.s. monetary policy setting,11
it is apparent that the actual length of the lags attached to
the economic variables is indeterminate from a theoretical
perspective. Here, appeal is made to a decision rule suggested by Hirotugu Akaike, which is based on the minimum
final prediction error (FPE) criterion. The selected lag structure discussed in the text follows directly from the results
of applying this criterion to the model above. 12
The estimation itself is carried out with data for the first
quarter of 1961 to the fourth quarter of 1983, using ordinary
least squares. In the final specification the lag length selections on the economic variables range between one and
four quarters. The significance of the FPE selected lags is
addressed in Table 1, with the results themselves presented
in Table 2.13
Table 1 shows each of the sets of lags to be significant at
the 1-percent level. Furthermore, the joint significance of
the set of interaction terms is confirmed by an F test:
F9 ,68

= 5.35 >


= 2.68.

In this way, the hypothesized modification of the central
bank's response to the economy due to the deficit is indeed
supported in the results.
Further testing was undertaken with respect to a possible
"critical threshold" in the effect of the deficit on central bank
policy. However, the results of a grid search test procedure
showed the effect to be continuous across the full range of
values for the deficit, with no particular lower bound apFederal Reserve Bank of Dallas

Table 1
Variabl e


Lag leng th


Test statistic

2 quarters

F2 .68


3 quarters

F3 .68

DC and (OfF. DC) .

2 quarters

F4 .68

OP and (OfF. OP) . ...

2 quarters

F4 .68

TB and (Off· TB) ..

4 quarters

F8 .68

UN and (Off. UN) .

1 quarter

F2 .68



= 11 .69
= 6.45
= 3.62
= 7.00
= 5.39
= 5.79

Critica l value
(05/ 01 )




= rate of growth of the monetary base.
= cyclically adjusted federal budget deficit divided


= rate of growth of


by trend GNP
real Federal Government purchases.

OP = rate of growth of the GN P price deflator .



three-month U.S. Treasury bill rate.

= unemployment rate.

parently required in order to bid forth a significant role for
government pressure.
Allowance for a possible reaction by the Federal Reserve
to international developments was also rejected. Neither
balance of payments nor exchange rate effects were found
to be significant in influencing monetary base growth. 14
In interpreting the results given in Table 2, the analysis focuses on the sign pattern for the sum of the lags on each
variable. The sign pattern observed in the results is listed
Policy response
at zero deficit

aOMB <


Effect of an increase
in the deficit


aOMB <0.


aOMB >0.


aOMB >0.


Economic Review - May 1986







The partial derivatives depicted in the first column measure the policy responses observed at a zero level of the
defi cit. These policy responses are interpreted as reflecting
the strategy pursued by the central bank when no pressure
is exerted by the government. The findings include a negative response to government purchases, a result that suggests a tendency by the Federal Reserve to offset increases
in federal spending by contracting monetary base growth.15
The standard precepts of countercyclical policy, meanwhile,
are satisfied by a contractionary movement of the monetary
base in response to inflation and by positive responses to
the interest rate and unemployment variables.
The second-column signs indicate how the Federal Reserve's stabilization policy is affected by an increase in the
deficit. In terms of government purchases, administration
pressure is, in fact, shown in the second column to induce
Federal Reserve policy to support rather than offset movements in federal spending.
The second-column's positive interaction between the
deficit and the response to inflation is consistent with the
postulated effect of the administration's shorter time horizon on the relative emphasis attached to price stability.

Table 2
quarterly lag

Constant .....


t statistic


-1 .62






-1 .54











UN: 1



Off· UN: 1



OMB: 1 ...

2 ...
Off: 1 ..

2 ..
OC: 1

OfF·OC: 1. . .

2 ..
OP: 1 ...

OfF·OP: 1 .

TB: 1 ..

2. . ,
Off· TB: 1 "

3 .. .


Dependent variable = OMB .
Sample period = 1961 :Q1-1983:Q4.
R2 = .63; DW = 1.97; SEE = .004.


is the coefficient of determination
adjusted for degrees of freedom
DW is the Durbin·Watson autocorrelation
test statistic
SEE is the standard error of the equation.

Here, the Federal Reserve is seen as having to "bend with the
wind" and reduce the size of the contraction that would
otherwise follow an increase in the inflation rate.
The negative sign on the interaction term for the Treasury
bill rate suggests that the administration also cares less
about persistently rising interest rates than does the Federal
Reserve. The implication is not that the administration
wants higher interest rates (and such an event would certainly increase its debt burden) but, rather, that it attaches
relatively less weight to interest rates than would be the case
for a member of the financial community.'6
The results described above certainly appear to be consistent with the theoretical model of Federal Reserve behavior. It is, however, in the effect of the deficit on the
response to unemployment that the findings appear somewhat implausible. Indeed, rather than the expansionary impetus that would be expected to follow from administration
focus on the full-employment objective, a negative effect is
observed instead. At the same time, it should be pOinted
out that the (one-quarter) interaction effect is insignificant
at the 5-percent level, leaving only weak opposition to the
ascribed role for the deficit in the Federal Reserve's feedback
The results are distinguished by strong evidence of stability in the model of monetary policy. A stable response pattern is, in fact, indicated throughout an extensive series of
Chow tests for structural change. Testing over the different
presidential administrations, and also over the tenure of the
different Federal Reserve chairmen, in each case provides
results consistent with continuity over the full sample. Two
further tests for potential breaks, coinciding with the transition to floating exchange rates in 1973 and with the October
1979 announced change in Federal Reserve operating procedures, once again provide test statistics that are insignificant at both the 5-percent and 10-percent levels.
The apparent robustness of the specification with respect
to sample-period changes certainly offers valuable support
for the model's applicability to the u.s. monetary policy
setting.'? There is strong empirical backing for the hypothesized effect of the federal budget deficit on central
bank stabilization policy.
The analysis has modeled monetary policy as a joint product of central bank trade-off weights and government pressure. In particular, the budget deficit is seen as the signal to
the monetary authority of the nature of the policy stance
sought by government. The demonstrable influence that
government pressure appears to exert on stabilization ob-


Federal Reserve Bank of Dallas

jectives is perhaps no more than the imposition of a
shorter-term outlook on the long-run planning horizon
otherwise relevant to the central bank. Such a perspective
is consistent with George Bach's observation that "in the
postwar world, whenever unemployment has risen substantially ..., Federal Reserve officials have moved solidly toward monetary ease, even when they still warned of the
danger of inflation.,,18
The empirical performance of the model is distinguished
in particular by its stability across regime changes. Such
stability has generally not been featured in other studies
that have modeled Federal Reserve policy over the same
post-1960 period. Indeed, the present framework, in which
the central bank is seen to "bend with the wind" represented
by fiscal pressure, is one that also may quite plausibly explain the general absence of sustained conflicts to be observed between the monetary and fiscal authorities .

stitution, ed leland B Yeager (Cambridge: Harvard University Press,

1962), 219-43.

SpeCifically, trend GNP (gross national product) is middle-expansion
trend GNP as computed by Frank de Leeuw and Thomas M . Holloway,
"Cyclical Adjustment of the Federal Budget and Federal Debt," Survey
of Current Business 63 (December 1983): 25-40.

7. The inclusion of the government purchases variable accords with certain optimal public finan ce considerations raised by Robert J. Barro, "On
the Determination of the Public Debt," Journal of Political Economy 87
(October 1979, pt. 1): 940-71 .
6. Interaction between the set of monetary policy coefficients and the
budget deficit has previously been suggested by Alan S. Blinder, "On the
Monetization of DefiCits," in The Econ o mic Consequences of Government Deficits, ed . laurence H. Meyer (Boston: Kluwer-Nijhoff, 1983),

Formal derivation of the signs on the interaction terms is provided by
Richard C. K. Burdekin, "Interaction Between Central Bank Behavior and
Fiscal Policymaking: The Case of the U.s.," Federal Reserve Bank of
Dallas Research Paper no. 8602 (Dallas, March 1986), app. A

The potential consequences of sustained central bank opposition to
government policy are illustrated by the "Coyne affair" in Canada,
where the forced resignation of the Governor of the Bank of Canada
was ultimately followed by amending the Bank of Canada Act in 1967
to grant the Minister of Finance the power to issue directives to the
bank. The Coyne crisis is discussed by J. W . O'Brien, Canadian Money
and Banking (New York: McGraw-Hili, 1964). 199-201.
2. This aspect of the literature is addressed by Thomas M. Havrilesky, who
further points to shifts in monetary policy that are specifically related to
th e red istributive component of fiscal policy Havrilesky's empirical
work suggests that increases in the ratio of social expenditures to aggregate income result in expanded rates of money supply growth. It is
postulated that a greater propensity for redistribution leads liberal parties to attempt more monetary surprises than is the case with conservative parties, implying that redistributive phenomena may explain why
money growth is often faster under left-of-center governments. See "A
Partisanship Theory of Fiscal and Monetary Regimes," Working Paper
Series, no. 86-08, Department of EconomiCS, Duke University (Durham,
N.C., 1986).

Richard H. Thaler and H. M . Shefrin, "An Economic Theory of SelfControl," Journal of Political Economy 89 (April 1981): 392-406.

4. By the same token, a country that has a perfectly "benevolent" central
bank (one that attaches the same weight to inflation as society does)
may suffer from having an inflation rate that is systematically too high.
In the presence of a distortion causing the market rate of employment
to be suboptimal, inflation arises because wage setters rationally fear
that the central bank will undertake expansionary policy aimed at raising employment systematically. Hence, a central bank with (known)
above-average commitment to fighting inflation is needed in order to
;"duce less inflationary wage bargains. See Kenneth Rogoff, "The Optimal Degree of Commitment to an Intermediate Monetary Target,"
Quarterly Journal of Economics 100 (November 1985): 1169-89.
5. These arguments are presented in Milton Friedman, "Should There Be
an Independent Monetary Authority?" in In Search of a Monetary Con-

Economic Review-May 1986

10. The importance to the Fed of support from its banking constituency has
been stressed by Neil T. Skaggs and Cheryl L Wasserkrug, "Banking
Sector Influence on the Relationship of Congress to the Federal Reserve
System," Public Choice 41 , no. 2 (1983): 295-306.
11 The results discussed in this section are taken from Richard C. K.
Burdekin, "The Interaction of Central Bank Behavior with Fiscal
Policymaking and the Political Busin ess Cycle: A Multi-Country Study"
(Ph.D. diss., University of Houston, 1985), chap. 3.
12. In applying Akaike's FPE criterion to the model, the maximum lag length
was set at six quarters Each variable was then tested in turn, initially
holding the lag length on the other variables at the maximum . The sixquarter limit received some justification in that in no case did the procedure choose a lag length of more than four for any variable. It may
be added that in order to correct for the large number of right-handside variables, the initial selections were used as the maxima in a second
application of the procedure. This correction actually had little effect
on the final specification, resulting only in the elimination of the third
and fourth lags on DP (both of which had a t statistic below 1). For details of the FPE criterion, see Hirotugu Akaike, "Statistical Predictor
Identification," Annals of the Institute of Statistical Mathematics 22 (1970):
13 The macroeconomic variables appearing on the right-hand side of the
monetary policy reaction function, by their very nature, tend to move
together over time. This tendency suggests the presence of multicollinearity, a phenomenon that generally makes it difficult to identify
the individual variables' contribution to the overall goodness of fit of the
equation. Multicollinearity leads to reduced values for the individual t
statistics, and the F statistics for the sets of variables depicted in Table
1 may also be understated. However, since each set is significant at the
1-percent level, the importance of each group of variables appears, in
any event, to be clearly demonstrated
14 Indeed, when the balance of payments deficit and the exchange rate
with the United Kingdom are added to the speCification, the FPE criterion in each case selects a lag length of zero.


15 It should be noted that because the deficit is held constant, this result

17. For results for Canada, France, the United Kingdom, and West Germany,

applies to the specific case of a balanced budget increase in government
16. As John T. Woolley puts it: "Even if bankers do not shape Federal Re·
serve decisions in an ongoing, detailed way,...the Federal Reserve is still
a central bank. It is socially and politically close to banks and banking"
(Monetary Politics: The Federal Reserve and the Politics of Monetary Policy
[Cambridge: Cambridge University Press, 1984], 87)

see Richard C K Burdekin, "Cross·Country Evidence on the Relationship
Between Central Banks and Governments," Federal Reserve Bank of
Dallas Research Paper no 8603 (Dallas, March 1986),
18. G L Bach, Making Monetary and Fiscal Policy (Washington, D .C:
Brookings Institution, 1971), 165

Formal Treatment of the Relationship
Between Central Bank and Government
The central bank and government are depicted as having
distinct loss functions related to the range of economic
variables embodied in the vector Y" which are themselves
associated with the corresponding set of target values Y"(.
The respective weights attached to deviations from the
targets can be represented by 0, and O2 below:


where LCB is the loss function of the central bank and
LGOV is the loss function of the government.

The longer view that has been ascribed to the central
bank would be expected to result in 0, not being equal to
O2, The problem faced by each policymaker is that of
minimizing the relevant loss function subject to the constraint represented by the structure of the economy. The
stochastic process determining the set of endogenous
variables for the system is represented by equation 3.

where Y, = a vector of economic variables
W, = a vector of policy instruments
X, = a vector of predetermined variables
(lagged values of the economic variables
and policy instruments)
A = a matrix of coefficients of the instruments
B = a matrix of coefficients of the predetermined
U, = a disturbance vector.
It can be shown that the solution to the policymaker's
problem collapses to a closed-loop rule of the form given
in (4) below:


W't = CXt

+ 6,

where C is a coefficient matrix and 6 is an error term containing Y",.
In this closed-loop rule, both C and e are a function not
only of the system's matrices given in (3) but also of the
relevant weightings of the trade-offs between target variables, as embodied in 0, and O2 , Therefore, a difference
in trade-off weights (that is, 0, #- O2 ) necessarily implies a
similar distinction between the respective optimal response
patterns-even were the perceived structure of the economy to be identical and both policymakers to use the same
The implications for monetary policy can now be addressed initially in terms of the decision rule that would be
selected by the government. The relevant monetary policy
instrument is taken to be the change in the monetary base:

where OMB, is the change in the monetary base.
Following from the arguments above, the coefficient
vector ag ov is dependent, first, on the perceived structure
of the economy and, second, on O2 from the government's
loss function. This is true also of the error term 6gov .
However, the central bank is, of course, the institution
entrusted with direct control over monetary policy. Were
the central bank itself able to set policy in an unconstrained
manner, then formulation of policy would be related to 0,
from the bank's own loss function (equation 1)-and not at
all to O2, The reaction function is written as

cb _




OMB( - aXt e .

To the extent that 0, is not equal to O2, it could be expected that agov is not equal to acb and 680V is not equal to
6 cb . However, given that the central bank is perceived to
face a constraint on the extent of government displeasure
it may engender, then both sets of coefficients will playa
role in the determination of monetary policy.
Federal Reserve Bank of Dallas

This proposition is examined in the context of an extended loss function for the central bank that allows for an
influence of pressure exerted by the government. The extent of the pressure is taken to depend on the gap between
the monetary policy that the government would have selected itself (from equation 5) and the monetary policy actually followed.
The extended loss function has the form

central bank (6) is instead a function of 0 1 from the central
bank's own loss function (1). This suggests that an increase
in OfF would be associated with an increase in pressure
because it implies an increase in the gap between actual
monetary policy and the policy that would have been set
by the government. 1
With the deficit serving as the proxy for government
pressure, we depart from the unconstrained feedback rule
(6) in favor of the general form given by


OMB, =




X, + e


0 1 = q(pa and
q(O) = 0 1 ,

q(oo) = O2 ,

where p, is the pressure exerted on the bank by the government.
In this framework the impact of pressure on the central
bank's loss function is embodied in the function q and implies an altered response pattern for the loss function. Indeed, as pressure increases, the central bank departs
further and further from the policy response consistent
with its unconstrained trade-off weights, such that in the
limit the observed trade-off weights approach the weights
represented in O2 .
In examining the influence of government pressure on
the central bank's feedback rule, the proxy for this pressure
is the cyclically adjusted deficit divided by trend GNP
(OfF). Here, the government's feedback rule for the deficit,
like the desired reaction function for monetary policy (5),
would be a function of the trade-off weights embodied in
O2, while the unconstFained policy reaction function for the

Economic Review -

May 1986

where IX def, rather than being a fixed parameter, is related
explicitly to the observed value of OfF (with a lag structure
corresponding to that of X,). In this way, the government's
loss function now affects monetary policy through the
indirect channel provided by the deficit, and this effect is
associated with central bank response to the implied
government pressure.
The nature of the dependence of IX def on the deficit can
be expressed most simply by the linear approximation

which provides the formal underpinning for the empirical
analysis discussed in the text.

1. In the model, minimum government pressure exists when the
deficit is zero. The effect of the deficit is further assumed to be
symmetric in that, just as a budget deficit implies pressure for
more expansionary policy, a budget surplus would be associated
with pressure for a more contractionary policy by the central


Energy's Contribution to the Growth
of Employment in Texas, 1972-1982
John K. Hill
Federal Reserve Bank of Dallas

Texas employment increased at an average annual rate of
5.2 percent from 1972 to 1982. In comparison, employment
in the United States rose only 2.3 percent per year. Much
of the rapid growth in Texas can be attributed to the prominent role of energy production in the state and to the fact
that energy production worldwide was encouraged by a
sharp rise in energy prices.
Chart 1 shows how Texas employment in two key energy
industries increased over the period. Employment in oil
and gas extraction, which includes engineers as well as
roughnecks, rose at an annual rate of 11.6 percent. Texas
was also an important supplier of machinery used in energy
extraction, and employment in oil field machinery manufacturing grew at a rate of 9.9 percent per year. The employment gains in these two industries alone accounted for 12
percent of the total rise in private nonagricultural employment over the 1972-82 period.
The influence of energy on Texas employment was not
limited to the direct gains in energy-producing industries.
As extraction employment increased, so did the demand for
professional, financial, and business services. The growing
oil field machinery industry increased the demand for primary and fabricated metal products. And the rise in state
personal income as a result of expansion in the energy sector increased the demand for health care, education, and
other consumer products.
Economic Review - May 1986

This article provides an estimate of energy's contribution
to employment growth in Texas during the 1972-82 period.
An input-output table is used to determine the employment
effects of growth in two energy-producing industries: oil
and gas extraction and oil field machinery manufacturing.
The effects include not only the intermediate demands
these industries make on other industries in the state but
also the additional employment gains arising as increases in
state personal income expand the demand for consumer
goods and services.
The results suggest that during the 1972-82 period, growth
in the two energy industries accounted for as much as 45
percent of the increase in total Texas employment. Energy,
then, would have contributed as much as 2 percentage
points to the annual growth rate of Texas employment.
Despite the importance of energy to the state's growth, the
gains in energy employment do leave unexplained a substantial portion of the rapid expansion that occurred in
Texas manufacturing.
The input-output model

The analysis is based on an input-output table that was
developed for the Texas economy.1 The table shows how
goods and services produced in each Texas industry during
1979 were distributed to other industries, households, and
governments. With this information, it is possible to mea11

Chart 1

Texas Employment in Energy-Producing Industries
(1972 =100)







--- -- ----

- -------- -- -- _-




_a _ ____ -- -





SOURCE OF PRIMARY DATA: US Bureau of Labor Statistics

sure the extent of economic interdependence among the
various sectors of the Texas economy_
The input-output table can be represented by the following system of equations:

(1 )





+ Cj Y + OJ,

i= 1, 2, ___ , N,








= state production in industry i
= output from industry i purchased by producers

in industry i, expressed as a fraction of Xj
Y = state personal income
cj = household demand for output from industry i,
expressed as a fraction of Y
OJ = autonomous demand for output from industry i
(including purchases by federal, state, and local
governments; capital formation; and exports
to out-of-state buyers)
Sj = income paid to state households by producers
in industry i, expressed as a fraction of X,
M = other state income (including government
transfer payments and income from out-of-state
The first set of equations divides industry outputs into their
intermediate and final uses. The intermediate demands on

industry i by industry i are written as aj A Final demands
are of two types. Household demand is expressed as a proportion of state personal income. The other category of
final demand, OJ, includes government expenditures and
exports. These purchases are not assumed to vary with
state income. The last equation details the contributions of
all industries to state personal income. Income not derived
from state production is denoted by M.
In addition to displaying information about the Texas
economy, the input-output table can be used to determine
the economy-wide impact of a given industrial expansion.
To do this, it is assumed that the expansion has a negligible
effect on the autonomous demands in other industries (OJ),
the income derived from out-of-state production (M), the
intermediate input coefficients (a j j), and the propensities to
consume (c;). Of particular note is the impliCit assumption
that there be no associated changes in relative prices,
changes that might lead to technical substitution among
factors of production or substitution among commodities in
household demand. This constraint requires the economy
in question to face highly elastic supplies of tradable commodities and primary factors of production. In view of this
requirement, input-output analysis is seen to be more
readily applicable to state economies than to national
The specific purpose of this article is to determine how
expansion in two energy-producing industries-oil and gas
extraction and oil field machinery-affected employment in
other state industries. The details of this procedure can now
Federal Reserve Bank of Dallas

be briefly sketched. Let indexes 1 and 2 denote the two
energy-producing industries. Given the above assumptions,
the equations in (1) can be rewritten to express the changes
in non-energy production and personal income in terms of
given changes in the two energy outputs.


IlXj -


ajjllXj - cjll Y = aj1 1lX1 + aj21lX2,


i= 3,4, ... , N;


sjllXj + Il Y

= s1 1lX1 + s21lX2·


This system of equations can be solved Simultaneously for
the changes in non-energy outputs. A conversion from
outputs to employment is made by using information on
industry labor-output ratios. 2 These ratios will be constant
and independent of changes in energy production, given
the earlier assumption of constant relative prices.
The solutions described above are intricate and complex.
To provide some understanding of the economic relationships involved, the accompanying box works through a
solution for a simple example involving one energy industry
and two non-energy industries.
Adjustments for induced construction

In static input-output models, like the one used here, there
is no means of analyzing capital formation. Changes in industry outputs do not induce construction of new homes,
buildings, or plants. This section outlines a simple method
for estimating the additional employment generated when
the construction sector is allowed to respond fully to the
energy expansion.
Suppose that each unit increase in total employment can
be associated with the employment of a additional construction workers. Ignore any differences in capital-labor
ratios across industries, and assume that the construction
induced by an economic expansion is independent of the
industry composition of that expansion. In practice, a will
be taken to be 0.08, the ratio of observed gains in construction employment to observed gains in total employment over the 1972-82 period. In addition to the direct
gains in construction employment, there will be indirect
gains of the kind summarized in the input-output model.
Let m denote the construction employment multiplier.
Then the change in employment attributable to the induced
gains in construction is given by maIlL', where ilL' is the

Economic Review - May 1986

total change in employment. If ilL is used to denote the
energy-related gains derived earlier from the input-output
model, ilL' can be expressed as


ilL' = AL

+ maAL'.

By solving equation 3 for ilL', it is possible to obtain a more
complete account of the effect of rising energy production
on state employment.
The above discussion has concentrated on changes in
total state employment. For a breakdown by industry, the
analogue to equation 3 is

where IlL j denotes the narrow estimate of energy-related
gains in industry i and mj is the construction multiplier for
industry i. Equation 4 can be evaluated using AL' from
equation 3 together with information on ALj and mj from the
input-output model.
Energy employment multipliers

To measure the dependence of the Texas economy on energy production, employment multipliers were derived for
each of two key energy industries: oil and gas extraction
and oil field machinery. In the original calculations, 46 individual Texas industries were identified. The results were
then aggregated into nine major sectors for easier interpretation. Table 1 shows the effects on private nonagricultural
Texas employment of an increase of 1,000 workers in each
of the two energy industries. The "unadjusted" columns
display the narrow estimates that were obtained directly
from the input-output table, while the "adjusted" columns
show data allowing for induced construction.
The figures in Table 1 represent the combined influence
of changes in intermediate and final demands on industry
employment. For some industries, particularly those in the
manufacturing sector, the effects are composed primarily
of changes in the intermediate demands made by other industries. But for other industries, such as retail trade and
many of the service industries, the effects are principally the
result of changes in consumer demand that are generated
by changes in state income. With the exception of the energy industries themselves, the largest employment gains
tend to be found in industries that cater more to household
demand than to business demand.
Regardless of the figures used, the total employment
multiplier for oil and gas extraction is larger than the multiplier for oil field machinery. Although many factors contribute to this result, the most compelling reason involves


A Simple Exercise in Input-Output Analysis
An example will serve to clarify the nature of the economic
interdependencies expressed in the input-output model.
Consider a state with three production sectors . First, there
is an energy-producing industry, denoted by the index E
Second, there are industries whose outputs are used as intermediate inputs by energy-producing firms. let the index
J represent a composite of these industries. The last group
to consider is composed of industries that produce consumer goods and services. A rise in energy employment
will increase state personal income, which, in turn, will expand the demand for final products. Use the index F to
represent state industries that produce final goods and
With this simple model in mind, it is possible to work
through the employment effects of a given increase in state
energy production let ~XE denote the change in energy
output and bj the amount of labor required per unit of
output in industry j. The most direct employment gains are
the bf~Xf new jobs created in the energy industry. Then
additional workers will be needed to produce the intermediate inputs required by energy-producing firms. The increase in intermediate production is given by d/f~XE' so
employment in sector I rises by bJilIE~Xf' The last source of
employment to consider is the final goods sector. An increase in state personal income expands consumer demand by CF~ Y, which raises employment in the final goods
sector by brCF~ y.
The changes in employment described above are totaled
in equation A.l:
(A. 1)




+ b,d'E~XE + bFCF~ Y

What remains is to relate the change in state income to the
change in state energy production. Begin with the incremental income generated in the energy and intermediate
goods sectors, SE~XE and Sld'f~Xf' respectively. These gains
are autonomous, being derived from production that is not
responsive to changes in state income. For ease of notation, denote the sum of these two gains by ~ YA- The process by which income is generated in the final goods sector
is more complicated. Through their effect on consumer


demand, the autonomous changes in income increase income in the final goods sector by the amount (SFCf)~ YABut this gain itself expands consumer spending, thus leading to a second round of income gains in the amount
(5FCF) 2 ~ YA- The process continues indefinitely, with income
gams of (5fCf)i~ YA in any ith round leading to additional
gains of (5rCF)i+ l~ YA in the next round.
When taken to the limit, the summation of all income
gains, including the autonomous ones, can be expressed

By substituting (A.2) into (A.l), the change in state employment can be expressed solely in terms of the change in
energy output And by substituting Mf/b Efor ~Xf and then
writing the result as a ratio of the change in energy employment, we have an algebraic rule for determining the
effect on total employment of a unit change in energy
(A.3) M/ME = 1

+ (b,jbE)aIE

+ (bF/bdcf[(SE + S,dIE)/(l

- SFCF)].

Equation A.3 reveals three elements that are crucial to the
size of the energy employment multiplier. First, there are
the terms (bdb E) and (bF/bd, which compare the labor requirements in energy production with those in other sectors of the economy. The more labor-intensive the energy
sector, the smaller will be its employment multiplier. Second, there is the parameter a'E, which measures the interdependence of energy and intermediate goods production.
The greater this interdependence, the larger are the employment gains in the intermediate goods sector and, given
the contribution intermediate goods production makes to
state income, the larger are the employment gains in the
final goods sector. The other important parameter is CF' the
propensity to consume state products. The larger the propensity to consume, the greater is the income generated
from a given energy expansion and, therefore, the greater
are the employment gains in the final goods sector.

Federal Reserve Bank of Dallas

Table 1
Employment generated by an increase of 1,000
workers in respective energy industry
Oil and gas extraction

Oil field machinery






Mining .
Nondurable manufacturing
Durable manufacturing . . .
. ........... .
Transportation and public utilities
Wholesale trade ...
Retail trade . .
Finance, insurance, and real estate ... ....... .





Total private nonagricultural employment ...





1 For induced construction
SOURCE OF PRIMARY DATA: Texas Department of Water Resources

the labor intensities of the two industries. The oil field machinery industry uses 2 1/2 times as much labor per given
value of output as does the extraction industry.3 As explained in the box, labor-intensive industries tend to have
smaller employment multipliers.
Finally, note the relatively small size of the construction
figures in the unadjusted columns. As discussed earlier, this
is due to an omitted linkage between capital stock and
output in static input-output models. Equations 3 and 4
were used to adjust the employment figures for a more
complete response in construction employment. 4 As can be
seen from Table 1, the adjustments had a significant effect
on the multipliers. The most dramatic revisions were, of
course, in the construction sector. The construction figures
were raised by a factor of 16 in the case of oil and gas extraction and 13 in the case of oil field machinery. The average effect of the revisions was to raise the extraction
multipliers by a factor of 1.33 and the machinery multipliers
by a factor of 1.29.
Estimates of energy-related growth

Reported in this section are estimates of the gains in Texas
employment from 1972 to 1982 that, either directly or indirectly, can be accounted for by growth in the state's oil and
gas extraction and oil field machinery industries. The estimates were obtained by multiplying the actual changes in
energy employment over this period by the appropriate
Economic Review - May 1986

employment multipliers and then adding up over the two
energy industries.
Chart 2 reveals the individual employment contributions
of the two energy industries. As can be seen, the bulk of the
energy-related growth was the result of gains in oil and gas
extraction. Extraction employment itself rose 195,000
workers. And when allowance is made for induced changes
in intermediate and final demands, the rise in extraction
employment ultimately was responsible for as many as
810,000 state jobs. In comparison, the oil field machinery
industry added only 44,000 workers to its payrolls, with a
total employment impact of 125,000 jobs.
Table 2 details the combined influence of the two energy
industries on Texas employment. Of the 2.1 million additional workers in the state, as many as 935,000, or 45 percent, were the direct or indirect result of expansion in
energy-producing industries. With the exception of mining
and construction, the relative contributions made by energy
ranged from 20 to 35 percent for the unadjusted figures and
26 to 43 percent when employment was adjusted for induced construction. s The largest percentage contributions
were in nondurable manufacturing, retail trade, and transportation and public utilities. Energy was less important to
the growth in wholesale trade, services, and durable
It is of interest to know how fast employment would have
risen in Texas without the growth in its energy industries and
whether the gains that were unrelated to energy were suffi15

Chart 2

Employment Contributions of Major
Texas Energy Industries
900r------------------------------------ - - -------,






SOURCES OF PRIMARY DATA: Texas Department of Water Resources
U.s Bureau of Labor Statistics

cient in themselves to allow the state to grow more rapidly
than the nation. These questions were addressed by calculating hypothetical growth rates, using only the employment gains that could not be accounted for by energy
growth. Selected figures for the construction-adjusted case
are shown in Chart 3. 6 The results suggest that without the
rise in energy production, total employment in Texas would

have increased at a rate of 3.2 percent per year, rather than
the rate of 5.2 percent actually recorded. The lower rate still
exceeds the national average but by a much smaller
margin? When broken down by sector, the results show
that employment in most sectors grew faster in Texas than
in the United States, even when the energy-related gains
were removed. An important exception is services. Service

Table 2
Change in employment
(Thousands of workers)
Unadjusted Adjusted'


......... .

Mining ..

Construction ... _ ........


Nondurable manufacturing
Durable manufacturing .
Transportation and public utilities


Wholesale trade
Retail trade ... .

Finance, insurance, and real estate
Services .




...... ...... .

. .... .. .


Total private nonagricultural employment



as percent of actual










ac tual


101 .5



1. For induced construction.
SOURCES OF PRIMARY DATA Texas Department of Water Resources _
U S Bureau of Labor Statistics


Federal Reserve Bank of Dallas

Chart 3

Employment Growth in Texas and the United States





















SOURCES OF PRIMARY DATA: Texas Department of Water Resources .
U.S Bureau of labor Statistics

employment would have grown at a slightly lower rate in
Texas had energy production in the state not increased. The
margins by which Texas outpaced the nation are largest in
durable and nondurable manufacturing. And this is true
whether or not energy-related gains are excluded.
Our analysis of the employment impact of Texas energy
growth suffers from two principal shortcomings. First, the
model fails to recognize any "crowding out" that may have
occurred as increases in energy production raised factor
prices in the state. Second, the model does not account for
any of the positive effects on industrial location of low rates
of business and personal taxation made possible by oil and
gas severance tax revenues. The first omission serves to
overstate the effect of energy growth on total state employment. The second results in employment estimates that
are downward biased.
Economic Review - May 1986

Input-output analysis is based on the assumption that the
supplies of factors of production are highly elastic. Ignored
is the possibility that a given industrial expansion may so
raise the prices of labor or land as to "crowd out" production
in other industries. The converse assumption would require
that factors be fixed in supply. Increased production in one
industry would then alter only the composition of employment, not its overall level. Except during periods of significant unemployment of resources, the second perspective
is the more appropriate one for national economies. But for
states such as Texas, which competes for labor in a national
market and for which constraints on the quantity of usable
land have never been very binding, the first perspective
would seem more accurate. The substantial increase in
Texas employment in recent years is evidence that the supply of labor to the state is highly elastic.
A second shortcoming of the analysis is that no attention
was paid to the effect of state and local taxes on industrial
location. Tax revenues from oil and gas production accounted for an average of 20 percent of total tax receipts in

Texas during the 1972-82 period. This contributed to the
fact that, excluding severance taxes, which strike only a
small subset of state industries, Texas ranked 47th lowest
among 48 U.s. states in effective rates of business taxation.8
It is likely, then, that gains in energy employment promoted
further economic growth by keeping taxes on other state
industries lower than they otherwise would have been.
The importance of this tax effect to Texas economic
growth is difficult to establish. But it likely wa not of the
same order of magnitude as the emp loyment effects identified by the input-output model. To a large extent, tax effort
in Texas has been light because the state provides a relatively low level of services. And an absence of services is in
itself an impediment to economic growth. Even so, studies
of the effect of taxes on business location suggest that state
and local taxes are only moderately important in decisions
about plant locations. Of much greater importance in these
decisions is the strength of local union activity.9
The rapid economic growth in Texas during the 19705 and
early 19805 is often attributed to the Important role of energy production in the state economy. Indeed, the results
presented here indicate that gains in energy employment
during this period were responsible for as much as 45 percent of the increase in state employment. But the analysis
leaves unexplained a significant portion of the state's
growth, particularly in manufacturing. After abstracting
from the direct and indirect effects of rising energy production, the annual rate of growth in Texas manufacturing
employment was found to be some 2 percentage points
greater than the national average.
It may be unreasonable to expect this rate of manufacturing growth to persist in coming years. The energy price
hikes of the 1970s hastened the obsolescence of energyinefficient plants, providing an enlarged pool of relocation
candidates. But the fact remains that many owners of
manufacturing facilities chose Texas as a location site. This
suggests that other locational attributes, such as low union
activity or a large supply of immigrant labor, also made
significant contributions.


Given the recent collapse in world oil prices and the expected layoffs in energy-producing industries, the near-term
outlook for the Texas economy is not encouraging. But the
results of this article provide some basis for optimism concerning the long-term prospects for growth in the state.

1. Mickey l Wright, Albert H. Glasscock, and Roy Easton, The Texas
Input-Output Model, 1979 (Austin: Texas Department of Water Resources, 1983) For a general discussion of input-output tables and their
use in economic analYSis, see Hollis B. Chenery and Paul G. Clark, Interindustry Economics (New York: John Wiley & Sons, 1959)

Data on ratios of employment to output are presented in Wright,
Glasscock, and Easton, Texas Input-Output Model.

3. In 1979 the number of employees per $1 million of output was 56 in
oil and gas extraction and 137 in oil field machinery. See Wright,
Glasscock, and Easton, Texas Input-Output Model.
4. The individual construction multipliers needed to make these adjustments came from the input-output model. They can be obtained from
the information presented in Table 1.

In the case of mining, the energy-related gains were more than 100
percent of the actual gains. This means that mining employment would
have fallen had employment in energy-producing industries remained


The actual growth rates in Chart 3 were derived by regressing the
natural logarithm of annual employment on a linear function of time.
The rates of non-energy growth were based on the level of employment
predicted for 1972 from the earlier regressions and on the change in
employment predicted for 1972-82, deflated by the ratio of energyrelated to actual employment gains.

7. The non-energy margin would be larger than the numbers in Chart 3
indicate had the effects of energy growth also been removed from the
national figures. But because the national labor market is more closed
than the Texas labor market, it would be misleading to make a similar
adjustment for the United States. With severe crowding out at the
national level, additional employment growth would have been
forthcoming in other sectors had energy production not increased .

The ranking was based on 1977 data. See William C. Wheaton, "Interstate Differences in the Level of Business Taxation: National Tax Journal
36 (March 1983): 83-94.

9. See Timothy J. Bartik, "Business Location Decisions in the United States:
Estimates of the Effects of Unionization, Taxes, and Other Characteristics of States," Journal of Business & Economic Statistics 3 Uanuary 1985):.

Federal Reserve Bank of Dallas