Download An empirical analysis of the determinants of the labor

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

Non-monetary economy wikipedia , lookup

Fei–Ranis model of economic growth wikipedia , lookup

Transcript
.
Journal of Policy Modelling
ELSEVIER Science Publishing Co
AUTHORS MANUSCRIPT TRANSMITTAL FORM
Name of Journal: Journal of Policy Models
Date Accepted:
Article Title: An empirical analysis of the
determinants of the labor force participation
rate in Puerto Rico
Journal Acronym: JPO
Name/Address of corresponding author:
Harri Ramcharran
University of Akron
Department of Finance, College of Business
302 Buchtel Common
Akron, Ohio
USA, 44325
Disk Enclosed:
Media Format: WordPerfect
Production Type: Hard Copy
Electronic version: Attached
Phone: 3309726882
E-mail: [email protected]
Publication Item Type:FLA
FLA Full Length Article
ECN Economic Note
SPI Special Issue
CNF Conference paper
Date Ms. Received:
Date Revised:
Number of Manuscript Pages:24
Number of Figures:0
Number of Tables:6
Editor's Notes:
Editorial Assistant: Sabah Cavallo
date:
7 Dreve Lansrode, Rhode St. Genese, Belgium 1640
[email protected]
1
An empirical analysis of the determinants of the labor force participation rate in Puerto Rico
Harri Ramcharran
Professor of Finance and International Business
Department of Finance
College of Business Administration
The University of Akron
Akron, OH 44325
USA
Phone: (330) 972-6882
Fax : (330) 972-5970
E-mail : [email protected]
2
An empirical analysis of the determinants of the labor force participation rate in Puerto Rico
Abstract:
This research empirically estimates the determinants of the labor force participation rate in Puerto Rico for
the period 1987-2008. The estimated results indicate statistically significant coefficients for the variables
real wages (labor income) and transfer payments (non-labor income). The implication of values of these
coefficients is that transfer payment considerations are more important than labor income in determining
labor supply choice and thus the LFPR. The variables indicating job opportunities and US- Puerto Rico
economic linkage are not significant. Policies aimed at reforming the transfer payment mechanism to
provide incentives to work are highly recommended.
JEL classification: J21, J22.
Key words: transfer payment, wage rate, labor force participation rate,
1. Introduction
Several empirical studies have focused on the labor supply response to welfare programs. Danziger, et al.
(1971) estimate that welfare recipients in the USA reduce hours of work by about one percent relative to total
work hours of all workers. Blau and Robins (1986) find lower labor force participation rate of all categories
(men, married women, single women and youths) of welfare recipients. Frish and Zussman (2008) report that
following the enactment of The Single Parent Family Law (which increased the income maintenance
allowance paid to single mothers) in Israel in 1992, the labor supply of uneducated single mothers fell by 10%
due to a decrease in the work hours. This study empirically examines the relationship between transfer
payments and the labor force participation rate (LFPR) in Puerto Rico using data for the period 1987-2008..
Several unique economic conditions attest to the importance of this study, they are: (a) transfer payments
from the US to individuals in Puerto Rico comprise about 21% of the Commonwealth personal income and
have increased steadily from $3,784.5 million in 1987 to $12, 252.8 million in 2008, (b) the residents bear no
tax burden for the benefits, and (c) the island’s LFPR averages about 45% from 1987 to 2008; it is the lowest
in the Latin America-Caribbean region and continues to be a topical issue of studies of Puerto Rico’s
economic development. Stagnant economic growth in recent years and the increasing percentage of the
island’s population living below the US poverty level have convinced policymakers that addressing the low
LFPR is crucial for promoting sustainable economic growth and raising the living standards of Puerto Ricans.
The formulation of effective policies thus requires an in depth knowledge of the determinants of the LFPR.
3
Additionally, the current economic recession offers no incentive for Puerto Ricans to migrate to the US
despite the absence of legal restrictions1.
The main focus of the investigation is to estimate a labor supply model that measures, inter alia, the
impact of labor income and non-labor income on the LFPR which is linked to workers’ choice between hours
worked and leisure. Ehrenberg and Smith (2006) and Borjas (2010) provide the theoretical rationale for the
methodology. We employ techniques of unit root, co-integration and regression analysis in a model using the
following independent variables: (i) average real wages and salaries, as a measure of labor income, (ii)
transfer payments (total and net) to individuals, as a measure of non-labor income, (iii) real GDP, economic
growth, as a measure of job opportunities, and (iv) the relative real GDP of the USA to that of real GDP of
Puerto Rico as the measure of economic linkage between both economies.
Several studies have documented possible explanations for the low LFPR in Puerto Rico. Bosworth
and Collins (2006) identify the following: (i) limited job opportunities, (ii) an underreported underground
economy, and (iii) a weak incentive to seek employment. Burtless and Sotomayor (2006) argue, from the
perspective of a labor supply model, that declining labor income and increasing federal (US government)
transfer payments (in the form of income supplement, disability payments and retirement benefits) to
individuals decrease the incentive to work thus lowering the LFPR. Finally, Enchautegui and Freeman (2006)
postulate the “rich uncle hypothesis”, contending that the link between a relatively poor Puerto Rico economy
and the rich US economy has created the conditions that reduce the incentive to supply labor in Puerto Rico.
These conditions are: (i) continuous federal government transfers to Puerto Ricans, and (ii) the free mobility
of labor between the US and Puerto Rico creates an incentive for Puerto Ricans to migrate to the US
The conclusions from the aforementioned studies are based on the analysis of data and descriptive
statistics obtained over a short time period, thus the implications must be interpreted with caution if used for
public policy decisions. Bosworth and Collins (2006) contend that the causes of the low LFPR are
fundamental and long-standing therefore a more analytical methodology is warranted for investigating this
issue with very significant social and economic ramifications.
4
The paper is organized as follows: (i) an overview of the LFPR and transfer payments pattern, (ii)
discussion of the model and data, (iii) tests of unit root and of cointegration, (iv) discussion of the regression
results, and (v) conclusion.
2. Transfer Payments, Wages, and LFPR in Puerto Rico
Until the mid-1970s, the LFPR in Puerto Rico was close to that of the mainland USA; also Puerto Rico
was able to generate wage increase and employment growth comparable to that of the USA Researchers
attempting to explain the significant drop in the LFPR of Puerto Rico often focus on two developments: (i)
the dramatic increase in USA social welfare transfer that started in the mid-1970s, and (ii) the slow growth
in real wages in Puerto Rico. Table A1 provides relevant data. Total transfer payments (TT) increased from
$3992.0 mil in 1987 to $13563.1 mil in 2008, an average of $7747.95 mil over this period. Transfer to
individuals (TI) was 94.8% of TT in 1987, this has decreased to 90.34% in 2008. Net transfer to individuals
(NTI) was 79.83% of TT in 1987 and has decreased to 76.87% in 2008. The main components of TI are: (i)
social security benefits, (ii) medicare benefits, (iii) veterans’ benefits, (iv) nutritional benefits, and (v)
housing assistance. NTI equals TI minus transfers from individuals to the federal government; these include
employees’ contribution to social security system and medicare contribution. Transfers to individuals
average about 21% of personal income over this period. Burtless and Sotomayor (2006) argue that the five
main transfer programs (Puerto Rico’s equivalent of food stamps, unemployment insurance, and social
security retirement and disability benefits, temporary assistance for needing families) have important work
discouraging effects because they reduce benefits to recipients who find employment. Furthermore, despite
average incomes are lower in Puerto Rico than in the USA, the social security programs in Puerto Rico offer
significant more attractive retirement and disability benefits.
Recent data indicate that wages in Puerto Rico average about 60% of that in the USA despite the
high educational attainment and skills of some workers in Puerto Rico and Puerto Ricans are subjected to
the US minimum wage law. Castillo-Freeman and Freeman (1992) show that efforts since 1974 to increase
the minimum wage in Puerto Rico to the US level have created a significant “unemployment effect” that
5
encouraged displaced worker to seek employment in the US Aggregate data show that total wages and
salaries as a percentage of Personal Income fell from 55.77% in 1987 to 48.06% in 2008, while total
workers’ compensation as a percentage of Personal Income fell from 64.8% to 55.45% over this period.
An analysis of LFPR must encompass different parameters of the economy. Davis and Rivera-Batiz
(2006) aver that job opportunities in Puerto Rico must be examined from these perspectives: (i) an
underdeveloped private sector, (ii) high public sector employment which discourages the emergence of
private sector employment, (iii) an industrial structure (partially created by Section 936 of the internal
revenue code) that is capital intensive and poorly connected with job opportunities, (iv) a minimum wage
law which discourages the hiring of less skilled workers and contributes to decreasing opportunities to
acquire experience and job training , and (v) large government transfers that hinder work incentives.2
3. Model
Several studies have utilized aggregate supply of labor models to analyze different characteristics of the
market.3 Wachter (1972) examines the determinants of the LFPR in USA within the framework of the
Phillips Curve using wages and inflation as the main independent variables. Kalachek et al (1979) utilize a
model of dynamic labor supply to measure adjustments to regional and demographic factors. There are also
empirical studies of the effects of different aspects of U.S transfer programs on labor supply, for example,
Fraker and Moffitt (1988) on the Food Stamp Program. The current literature shows no such study of Puerto
Rico. Kwok (2009) analysis of the labor force response (work-leisure choice) to changes in wealth (nonlabor income) stimulates new research interest in this topic.
The methodology utilizes a model with independent variables aimed to integrate and analyze two
dimensions of the determinants of the LFPR: (i) a labor supply model that estimates the impact of labor
income (real wages/salaries) and non-labor income (total and net transfer payments to individuals), and (ii) a
macroeconomic aspect that examines the impact of job opportunities (indicated by real economic growth),
and the incentives for Puerto Ricans to migrate to the USA (indicated by the economic linkage).
6
The labor supply model focuses on a worker’s choice between hours worked and time spent on
leisure whose price (opportunity cost) is the real wage rate. It hypothesizes, a lá Ehrenberg and Smith
(2006) and Borjas (2010), the following: (i) an increase in non-labor income (government subsidies and
transfer payments), holding real labor income (wage rate) constant, will decrease hours worked or increase
leisure time, referred as the income effect, and (ii) an increasing real labor income, holding non-labor
income constant, will increase hours worked or decrease leisure time, referred as the substitution effect. The
presence of both effects (income and the substitution) working in opposite directions, could create
ambiguity in overall labor supply response, thus the actual labor supply is the sum of both effects. If the
income effect dominates the substitution effect, the number of hours worked decreases while if the
substitution effect dominates the income effect, the number of hours worked increases.4
The dependent variable is the LFPR. The independent variables are: (i) for labor income, we use
total wages and salaries divided by number of employment to get average real wage and salaries (AWS)
since systematic data on wages are not published on a regular/systemic basis (ii) transfers to individuals
(TI), (iii) net transfers to individuals (NTI), (iv) GDPR is the real GDP, (v) GNPR is the real GNP and (vi)
RELGDPR is the real GDP of the US to relative GDP of Puerto Rico; this variable captures the link between
the economies of both countries. Many researchers have argued that the growing divergence between the
GDP and the GNP of Puerto Rico has created problems in choosing which of the two variables is useful for
assessing macroeconomic performance.5 Enchautegui and Freeman (2006) suggest the use of GNP rather
than GDP. We estimate the impact of both variables in separate models. The relative performance
(economic linkage) is an indicator of the potential for migration to the US or increasing federal transfers to
Puerto Rico. Descriptive statistics of the data are presented on Table A2.
We estimate the following models with the inclusion of time (1987-2008) as an independent
variable; this coefficient measures any significant secular trend or changes in the LFPR not explained by the
model.
Model 1.
LFPR = a0 + a1AWS + a2TI + a5TIME
Model 2.
LFPR = a0 + a1AWS +a2TI + a3GDPR + a5TIME
Eq.1
Eq.2
7
Model 3.
LFPR = a0 + a1AWS + a2TI + a3GNPR + a5TIME
Eq.3
Model 4.
LFPR = a0 + a1AWS + a2TI + a3GDPR + a4RELGDPR + a5TIME
Eq.4
Model 5.
LFPR = a0 + a1AWS + a2TI + a3GNPR + a4RELGDPR + a5TIME
Eq.5
We also re-estimate Model 1-5 substituting NTI for TI.
We hypothesize the coefficient of (i) AWS to be positive, a lá the substitution effect of the labor
supply model, (ii) TI and NTI to be negative, a lá the income effect of the labor supply model, (iii) GDPR
and GNPR to be positive since economic growth creates increasing demand for labor and increasing
participation rate, and (iv) RELGDPR to be negative assuming that a faster growth in GDPR of the USA
relative to GDPR of Puerto Rico could create an incentive for migration to the USA and thus lower the
LFPR. This higher relative ratio could also cause an increase in transfer payments to PR and lower the
LFPR.
4. Data
The data sources include various issues of the following: (i) Informe Económico al Gobernador y a la
Asamblea Legislativa (Economic Report to the Governor and to the Legislative Assembly) published by
Estado Libre Asociado de Puerto Rico, Oficina del Gobernador, Junta de Planificación (Commonwealth of
Puerto Rico, Office of the Governor, Planning Board), (ii) Ingreso y Producto (Income and Product)
published by Estado Libre Asociado de Puerto Rico, Oficina del Gobernador, Junta de Planificación
(Commonwealth of Puerto Rico, Office of the Governor, Planning Board)
5. Tests of stationarity and of cointegration
The analysis of the regression results of the model is preceded by a discussion of the rationale for using
(a) test of unit root, and (b) test of cointegration.
(a) Test of Unit Root
Test of stationarity (or non-stationarity) of data is important for empirical studies using time series
data to avoid the problems of “spurious regression”. There are several tests discussed in the literature, for
example, Gujarati and Porter (2009 Ch.21) and Enders (1995). The unit root test is very prominent; we
conduct the ADF (Augmented Dickey-Fuller) test which corrects for uncorrelated error terms. The results,
8
shown on Table 1, indicate that the null hypothesis of the existence of unit root (non-stationarity of the data)
is rejected at the first difference level in most cases.
[Table 1 here]
(b) Test of co-integration
The importance of a long run stable relationship among the variables used in time series model is
widely documented in the literature on econometric studies, (Enders 1995). Such relationship is crucial for
statistical inferences that are used for policy making. If a cointegrating relation among variables does exist it
indicates a long run stable model over the period analyzed and that the results are valid. Even if a model
uses some variables that are non-stationary (presence of unit root), it is important that there exists a
combination that is cointegrated. Models with non-stationary data that are not cointegrated yield results that
are not valid – ‘spurious regression’ problem (Engle and Granger, 1987). A variety of methods for testing
cointegration have been proposed in the literature, (Maddala and Kim, 1998); we utilize the techniques
developed by Johansen (1988), they are (i) the “trace” test and (ii) the “maximum eigenvalue” test. These
tests are applied to four series (based on four models) with the four conventional trend assumptions. The
results, presented on Table 2, indicate that either of the two tests shows that the variables are cointegrated.
[Table 2 here]
6. Results
With evidence that the variables used in the models are stationary and co-integrated, we infer that the
regression results are reliable. The estimated results of the five models are presented on Table 3. The R2
values are about 0.70; and the F statistics for all five models are significant. Since R2 < DW there is no
reason to suspect that the estimated results are spurious, a lá Granger and Newbold (1974). This is
consistent with the conclusion of the Johansen’s tests that the variables are co-integrated and lends support
for the reliability of the estimated results. The relatively low R2 could be explained by the insignificance of
two independent variables (discussed later) and /or possible exclusion of other independent variable in the
model. Previous studies mention the existence of an unrecorded underground economy in Puerto Rico
9
which employs “informally” a large population. With no published data, this variable could not be included
in the model.
[Table 3 here]
The coefficient of the variable AWS is positive and statistically significant at the 1% level; it indicates
that in increase in real wages (holding TI constant) increases the LFPR; this is consistent with the
substitution effect of the labor supply hypothesis. The coefficient of the variable TI is negative and
statistically significant, indicating that an increase in TI (holding AWS constant) decreases LFPR, this is
consistent with the income effect of the labor supply hypothesis. For all five models, based on the values of
the coefficients, the impact (negative) of TI on LFPR exceeds that of the small (positive) impact of AWS.
Together the results indicate that the income effect dominates the substitution effect and could result in a
decrease (increase) in workers’ choice for work (leisure). The policy implications are important, transfer
payment (non-labor income) considerations are more important than real wages (labor income) in impacting
decisions to work and thus the LFPR in Puerto Rico. The results are similar to those of other studies
indicating the disincentive work effects of welfare programs.
Neither measurement of economic growth, GDPR nor GNPR, is significant (Eq. 2 and Eq. 3); this
suggests the limitation of this variable as a policy instrument to stimulate the LFPR. The economy’s largest
and dynamic sector, manufacturing, employs only a small portion of the labor force. 6 This sector is capital
intensive and poorly connected to job opportunities and incentives to work. We extend the analysis of the
implications of economic growth on LFPR by examining labor supply response to business cycle in-terms
of the (a) added worker effect- AWE, and (b) discouraged worker effect- DWE.7 Following Borjas (2010)
we find a negative correlation between LFPR and the unemployment rate; this suggests the dominance of
DWE over AWE indicating that the high level of unemployment convinces many workers to give up job
search and drop out of the labor market. We infer that the DWE is reinforced by the availability of transfer
payments.
The variable, RELGDPR, linking the relative performance of the US economy and the Puerto Rico
economy, is also not significant (Eq. 4 and Eq. 5). Recent data indicates a highly positive correlation
10
between these two economies; only a countercyclical (negative correlation) economic performance could
possibly encourage migration from Puerto Rico to the USA and lower the LFPR.7 Additionally, despite the
relatively higher wage rate in the USA, Puerto Ricans, with low level skills, are unlikely to migrate to the
USA because of the generous disability, unemployment and social security benefits available on the island.
Borjas (2008) further contends that because of the high migration cost many Puerto Ricans choose not to
move to the USA
[Table 4 here]
The results of the models using NTI rather than TI are presented on Table 4. Like the results of the
previous models, only AWS and NTI are statistically significant, and the implications of their significance
are the same as discussed earlier.
7. Conclusion
Labor is the most abundant factor of production in any country. A country’s economic well being in the
long run depends, inter alia, primarily on the willingness of its people to work. The LFPR in Puerto Rico is
among the lowest in the region despite the island’s relatively more stable economy. The correlation
coefficient (computed from recent data) between the unemployment rate and the LFPR in Puerto Rico is 0.33, this statistic indicates that an increase in LFPR is important to decrease the unemployment rate.
The low LFPR in recent years emerges as a critical component of any strategy/policy aimed at
raising the living standards of Puerto Ricans. Recent studies have listed the following causes: (i) low wage
rate, (ii) federal transfer payments to individuals, and (iii) the industrial structure of the economy that is
poorly connected to job creation. The findings of this research are consistent with labor supply theory; they
include: (i) a positive relationship between LFPR and real wage, and (ii) a negative relationship between
LFPR and transfer payments to individuals. The implication of the estimated coefficients indicates that
transfer payment (non-labor income) considerations are more important than labor income in determining
the LFPR.
11
The policy ramifications of the results strongly suggest that reforming the transfer payment
mechanism and wage rate structure are important for increasing the LFPR. Policy recommendations must
consider both variables jointly, not separately, since both impacts the LFPR in opposite ways. Public efforts
should help improve a transfer recipient’s job search behavior; such efforts could lessen the work
disincentive effects of the transfer programs. Several recommendations have been proposed; they include
the following8: (a) collaboration between the government of Puerto Rico and USA to redesign the transfer
programs to encourage work, (b) imposed an earned income tax credit as in the United States, (c) give tax
credit to firms on the basis of number of job created, (d) discourage the rate of withdraw of older men from
the work force,(e) modify the benefits schedule to reflect new eligibility rules for assistance, (f) a wage rate
policy aimed at convergence with that of the USA since wage rates across industries average about 50-65%
of that in the USA; any decision on wage increases should consider not only the positive supply aspect of
the labor market but also the negative demand side.9
From the macroeconomic perspective, there are many criticism of the poor job creating aspect of the
current industrial structure of the economy, especially the negative impact of the US tax policy which,
according to Bosworth and Collins (2006) , has done a disservice to Puerto Rico by providing US
companies with investments that generate little employment or local economy linkages. Tax reform should
aim at creating production incentives to firms. Other recommendations include the expansion of the private
sector since the public sector is the largest employer on the island and public sector employment varies
significantly with fiscal crisis, and enlarging the industrial base through the clustering approach.10
Recommendation for future research on this topic should take consideration the following: (a) an
improvement in the measurement of LFPR because there are many Puerto Ricans who work in the informal
sector or are self-employed and are not included in the standard labor force survey, (b) estimating the value
of the “underground economy” since those who work “illegally” may not report their working activities;
there is also evidence of a number of “undocumented” immigrants in Puerto Rico receiving transfer
payments and working in the informal sector, and (c) estimating, given the availability of published data, the
12
determinants of LFPR for male and for female separately since in recent years the data on the LFPR shows
an increasing trend for female and a decreasing trend for male.
13
Notes
1. Puerto Ricans are classified as citizens of the US under the Jones Act of 1917, they could move freely to the
U.S without legal restrictions.
2. Section 936 of the Tax Code. Under the US Tax Code, profits earned by US corporations in Puerto Rico are
exempted from US taxation. The provisions were implemented in 1976 and repealed in 1995 with a five
year phase out.
3. Other approaches to analyze labor force participation rate focus on independent variables that include age,
gender, race, marital status, level of education/skills.
4. If the substitution effect dominates, the worker’s supply curve will be positively sloped, if the income effect
dominates, the labor supply curve will be negatively sloped. It is possible that an individual supply curve of
labor could be positively sloped in some ranges of wages and negatively sloped in others. A negatively
sloped supply curve suggests that work is an inferior good and leisure a normal good.
5. The gap between GDP and GNP has increased significantly in recent years reflecting the increasing payment to
foreign factors of production. The nominal GDP/GNP ratio increased from 1.3047% in 1980 to 1.5343% in
2008, the real GDP/GNP ratio increased from 1.1635% to 1.5974% over this period. The main reason for
this gap is US Tax Code Section 936 which creates the incentives to use transfer pricing to shift reported
income to Puerto Rico to avoid US taxation. This overstates GDP, the production of goods and services on
the island.
6. Manufacturing employment as a percentage of total employment was 15.24 % in 1998; it has decreased to
10.74% in 2006.
7. One popular method of analyzing the labor supply response to business cycle is discussing the added worker
effect (AWE) and the discouraged worker effect (DWE). The AWE implies that the labor force participation
rate of the secondary workers has a counter-cyclical trend, it increases during recession and decreases
during expansion. The DWE implies that many unemployed workers, rather than incurring the costs of
fruitless job search activities, drop out of the labor force. If the DWE dominates AWE the correlation
between the participation rate and the unemployment rate is negative; if the AWE dominates the correlation
is positive. For an elaborate discussion see: Borjas ( 2010), Chapter 2.
8. For an excellent discussion on reform proposals, see Enchautegui and Freeman (2006) and Burtless and
Sotomayor (2006)
9. Castillo-Freeman and Freeman (1992) show that efforts since 1974 to increase the minimum wage in Puerto
Rico to the US level have decreased employment by 9% and increased unemployment by 3 %. This issue is
discussed in Borjas (2010) page 128.
10.
The cluster approach adopted by the Puerto Rico Industrial Development Company (PRIDCO) identifies
eight strategic clusters to become the focus of the government’s promotion program: (i) pharmaceutical
14
products, (ii) biotechnology, (iii) plastics, (iv) electronic and communications products, (v) medical
instruments and devices, (vi) personal and health care products, (vii) optical products and (viii) construction
services and materials. For a detailed elaboration, see Lawrence and Lara ( 2006)
15
References
Blau, D. M and P.K Robins (1986) “Labor supply response to Welfare Programs; A dynamic analysis”. Journal of Labor
Economics. 4,82-104.
Borjas, G (2010) Labor Economics (5th Edition) McGraw–Hill Irwin
Borjas, G (2008) “Labor Outflows and Labor Inflows in Puerto Rico” Journal of Human Capital, 2: 32-68
Bosworth, B. and S. M. Collins (2006) “Economic Growth” Chapter 2 in The Economy of Puerto Rico, Brookings
Institution Press, Washington, D.C.
Burtless, G. and O. Sotomayor (2006) “Labor Supply and Public Transfers” Chapter 3 in The Economy of Puerto Rico,
Brookings Institution Press, Washington, D.C.
Castillo-Freeman, A.J and R.B Freeman (1992) “When the minimum wage really bites: The Effect of the US Level
minimum on Puerto Rico” in George Borjas and R.B Freeman, Immigration in the work force: Economic Consequences
for the United States and Source Areas: University of Chicago Press, (1992) pp 177-211.
Collins,S, B.P. Bosworth, and Miguel A. Soto-Class (2006) The Economy of Puerto Rico, Brookings Institution Press,
Washington, D.C.
Danziger, Sheldon, R. Haveman and R. Plotnick (1981) “How Income transfers affect Work, Savings, and the Income
Distribution.” Journal of Economic Literature. 14, 975-1020
Davis S. J. and L. A. Rivera-Batiz (2006) “The Climate for Business Development and Employment Growth” Chapter 6
in The Economy of Puerto Rico, Brookings Institution Press, Washington, D.C.
Economic Report to the Governor and to the Legislative Assembly, published by Commonwealth of Puerto Rico, Office of
the Governor, Planning Board (various issues).
Ehrenberg, R. G. and R. S. Smith (2006) Modern Labor Economics: Theory and Public Policy. Pearson Education Inc.
Enchautegui, M. E. and R. B. Freeman (2006) Chapter 4 “The Rich Uncle (Sam) Hypothesis” in The Economy of Puerto
Rico, Brookings Institution Press, Washington, D.C.
Enders, W. 1995. Applied Econometric Time Series, John Wiley & Sons.
Engle, R. F. and Granger, C. W. J (1987) Co-integration and error correction: representation, estimation, and testing,
Econometrica, 55, 251-76.
Fraker, T and R. Moffitt (1988) “The effect of food stamps on labor supply: A bivariate selection model”, Journal of
Public Economics, 35, 25-56.
Frish, Roni and Noam Zussman (2008)
Journal of Socio-Economics 37, 627-643.
“ The effects of transfer payments on the labor supply of single mothers”.
Granger, C. W. J. and P. Newbold (1974) “Spurious Regressions in Econometrics,” Journal of Econometrics, vol.2, pp.
111-120.
Gujarati, D. N. and D. C. Porter (2009) Basic Econometrics (5th edition), McGraw-Hill Irwin.
Income and Product, published by Commonwealth of Puerto Rico, Office of the Governor, Planning Board (various
issues).
16
Johansen, S (1988) Statistical analysis of co-integrating vectors, Journal of Economic Dynamics and Control, 12, 231-54.
Kalachek, E.D, F.Q.Raines and D. Larson. “The determination of labor supply; A dynamic model.” Industrial and Labor
Relations Review,32, 367-377.
Kwok, J (2009) “Labor supply responses to changes in wealth and credit.” FRSB Economic Letter. Number 2009-05,
January 30th.
Lawrence, R. Z. and J. Lara (2006) “Trade Performance and Industrial Policy” Chapter 9 in the Economy of Puerto Rico,
Brookings Institution Press, Washington, D.C.
Maddala G. S. and I. M. Kim (1998) Unit Roots, Cointegration, and Structural Change, Cambridge University Press.
Wachter, M.L (1972) ”A labor supply model for secondary workers.” The Review of Economics and Statistics.54,141-151.
17
Table A1 Transfers, Personal Income, GDP R and GNPR ($ mil)
Year
TT
TI
PI
NTI
PIN
GNPR
GDPR
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
3992.0
4073.0
4288.8
4870.9
4973.2
5107.8
5477.7
5956.9
6236.4
6803.8
7399.0
7758.3
8627.2
8659.4
9316.9
9818.2
10451.0
10086.6
10551.0
11210.4
11827.3
13563.1
3784.5
3840.6
4082.3
4648.5
4708.5
4903.4
5279.0
5630.4
5911.6
6419.4
6942.8
7175.3
7866.2
7868.2
8421.6
8918.5
9618.9
9161.0
9546.7
10065.7
10527.4
12252.8
597.6
691.9
765.5
817.4
863.8
917.9
980.0
1002.6
1051.7
1129.0
1158.4
1231.0
1237.1
1326.1
1407.9
1456.0
1547.7
1700.2
1791.5
1868.6
1804.5
1827.2
3186.9
3148.7
3316.8
3831.1
3844.7
3985.5
4299.0
4627.8
4859.9
5290.4
5784.4
5944.3
6629.1
6542.1
7013.7
7462.5
8071.2
7460.8
7755.2
8197.1
8722.9
10425.6
16465.0
17755.7
19198.0
21105.0
21883.9
22910.2
24612.4
25863.5
27377.6
29914.1
32663.3
34340.2
36614.5
38855.7
41079.5
42038.6
44215.6
45565.9
48820.2
50842.3
52294.7
56201.4
4428.8
4625.6
4807.7
4929.8
4972.8
5011.5
5177.8
5308.9
6491.8
5671.2
5864.2
6054.7
6300.1
6487.1
6585.1
6562.6
6702.7
6886.2
7019.6
7055.5
6918.8
6742.4
5705.2
6076.5
6376.7
6617.8
6770.3
7079.3
7408.1
7718.2
8069.3
8256.0
8658.9
9137.8
9630.3
9945.4
10573.3
10670.2
10675.9
10998.8
11089.7
11073.2
10920.0
10770.0
TT = Total Transfers, TI = Transfers to Individuals, PI = Payments by Individuals
NTI = Net Transfers by Individuals (TI - PI), PIN = Personal Income
18
Table A1(continued) : TI/TT, NTI/TT, TI/PIN (%)
Year
TI/TT
NTI/TT
TI/PIN
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
94.80%
94.29%
95.19%
95.43%
94.68%
96.00%
96.37%
94.52%
94.79%
94.35%
93.83%
92.49%
91.18%
90.86%
90.39%
90.84%
92.04%
90.82%
90.48%
89.79%
89.01%
90.34%
79.83%
77.31%
77.34%
78.65%
77.31%
78.03%
78.48%
77.69%
77.93%
77.76%
78.18%
76.62%
76.84%
75.55%
75.28%
76.01%
77.23%
73.97%
73.50%
73.12%
73.75%
76.87%
22.99%
21.63%
21.26%
22.03%
21.52%
21.40%
21.45%
21.77%
21.59%
21.46%
21.26%
20.89%
21.48%
20.25%
20.50%
21.22%
21.75%
20.10%
19.55%
19.80%
20.13%
21.80%
19
Table A2 Descriptive Statistics
AWS
Mean
Median
Maximum
Minimum
Std. Dev.
Skewness
Kurtosis
16078.52
15776.52
22192.60
10652.09
3539.936
0.081485
1.752290
GDPR
8828.223
8898.350
11089.70
5705.200
1884.105
-0.192113
1.533271
GDPUSA
8953964.
8885200.
11727400
6475100.
1700723.
0.140254
1.659862
GNPR
LFPR
5936.586
6177.400
7055.500
4428.800
878.5198
-0.300228
1.578569
46.29545
46.20000
48.10000
43.90000
1.058965
-0.124765
2.687587
20
NTI
5927.259
5864.350
10425.60
3148.700
2051.150
0.317668
2.165489
RELGDP
1021.420
1012.014
1134.947
935.4412
53.74460
0.452496
2.414797
TI
7162.423
7059.050
12252.80
3784.500
2434.634
0.269112
2.034596
Table 1 Results of ADF Test for Unit Roots
Variable
Test in
Included in Test
Coefficient
t (tau) Value
Prob
Decision*
LFPR
1st Difference
Constant
Constant & Trend
None
-1.07555
-1.147923
-1.0516
-3.4962
-3.7918
-3.670549
0.0199
0.0403
0.0010
Reject Ho
Reject Ho
Reject Ho
TI
1st Difference
Constant
Constant & Trend
None
-1.19457
-1.2802
0.00103
-3.4585
-3.8124
0.003527
0.0208
0.0376
0.6713
Reject Ho
Reject Ho
Don’t Reject Ho
NTI
1st Difference
Constant
Constant & Trend
None
-1.0647
-1.13678
-0.410682
-3.1589
-3.43117
-1.4381
0.0381
0.0754
0.136
Reject Ho
Reject Ho
Don’t Reject Ho
AWS
1st Difference
Constant
Constant & Trend
None
-1.467024
-1.505886
-0.048686
-5.8459
-6.05864
-0.42164
0.000
0.0001
0.5234
Reject Ho
Reject Ho
Don’t Reject Ho
GDPR
1st Difference
Constant
Constant & Trend
None
-0.4191
-0.5727
-0.2064
-1.9037
-2.4604
-1.6066
0.324
0.3414
0.1002
Don’t Reject Ho
Don’t Reject Ho
Don’t Reject Ho
GNPR
1st Difference
Constant
Constant & Trend
None
-1.428
-1.4573
-1.268
-6.56
-6.6743
-5.747
0.000
0.0001
0.000
Reject Ho
Reject Ho
Reject Ho
* Ho: unit root exists. Decision is based on the Augmented Dickey-Fuller test statistic, MacKinnon (1996).
21
Table 2 Johansen Cointegration
Rank Test: Trace and Maximum Eigenvalue
1. Series: LFPR, AWS, TI
Trend Assumption
Trace Test (0.05 level)
Maximum Eigenvalue (0.05)
(i) No deterministic trend
(ii) No deterministic trend (restricted constant)
(iii)Linear deterministic trend
(iv) Linear deterministic trend (restricted)
2 cointegrating eqn(s)
2 cointegrating eqn(s)
no cointegration
1 cointegrating eqn(s)
2 cointegrating eqn(s)
2 cointegrating eqn(s)
1 cointegrating eqn(s)
1 cointegrating eqn(s)
2. Series: LFPR, AWS, NTI
Trend Assumption
Trace Test (0.05 level)
Maximum Eigenvalue (0.05)
(i) No deterministic trend
(ii) No deterministic trend (restricted constant)
(iii)Linear deterministic trend
(iv) Linear deterministic trend (restricted)
2 cointegrating eqn(s)
2 cointegrating eqn(s)
no cointegration
1 cointegrating eqn(s)
2 cointegrating eqn(s)
no cointegration
no cointegration
1 cointegrating eqn(s)
3. Series: LFPR, AWS, TI, GDPR, RELGDP
Trend Assumption
Trace Test (0.05 level)
Maximum Eigenvalue (0.05)
(i) No deterministic trend
(ii) No deterministic trend (restricted constant)
(iii)Linear deterministic trend
(iv) Linear deterministic trend (restricted)
4 cointegrating eqn(s)
5 cointegrating eqn(s)
3 cointegrating eqn(s)
3 cointegrating eqn(s)
no cointegration
no cointegration
no cointegration
1 cointegrating eqn(s)
4. Series: LFPR, AWS, NTI, GDPR, RELGDP
Trend Assumption
Trace Test (0.05 level)
Maximum Eigenvalue (0.05)
(i) No deterministic trend
(ii) No deterministic trend (restricted constant)
(iii)Linear deterministic trend
(iv) Linear deterministic trend (restricted)
4 cointegrating eqn(s)
4 cointegrating eqn(s)
3 cointegrating eqn(s)
3 cointegrating eqn(s)
no cointegration
no cointegration
no cointegration
1 cointegrating eqn(s)
22
Table 3: Regression Results: Dependent Variable, LFPR
Variable
Model 1
Model 2
Model 3
Model 4
Model 5
Constant
-2820.12
-2846.5
-3053.078
-2808.15
-2970.07
AWS
0.00171
(3.522)***
0.00169
(3.29)***
0.001686
(3.43)***
0.00158
(2.78)**
0.00166
(3.309)***
TI
-0.0122
(-3.095)***
-0.0124
(-2.749)**
-0.014
(-3.082)***
-0.0115
(-2.358)**
-0.0128
(-2.609)**
GDPR
-4.51E-05
(-0.10595)
GNPR
-0.00035
(-0.462)
-0.000507
(-0.8195)
RELGDPR
-0.0007
(-1.004)
-0.0033
(-0.492)
-0.00268
(-0.634)
YEAR
1.453
(5.299)***
1.466
(4.74)***
1.5717
(5.03)***
1.449
(4.555)***
1.531
(4.72)***
R2
0.683
0.683
0.69
0.688
0.702
DW
1.549
1.55
1.53
1.48
1.479
F
12.95
9.18
9.7
7.06
7.57
Prob F
0.000
0.000
0.000
0.001
0.000
Note: LFPR = Labor Force Participation Rate, TI = Transfers to Individuals, NTI = Net Transfers to Individuals,
GNPR = Real GNP, GDPR = Real GDP, RELGDPR = (GDPUSA/GDPPR)
* denotes significance at 10% level, ** denotes significance at 5% level, *** denotes significance at 1% level
23
Table 4: Regression Results: Dependent Variable, LFPR
Variable
Model 2
Model 5
Model 6
Model 9
Model 10
Constant
-2690.17
-2851.4
-2674.58
-2648.3
-2776.99
AWS
0.00176
(3.66)***
0.00175
(3.5767)***
0.00177
(3.49)***
0.0016
(2.86)**
0.0017
(3.43)***
NTI
-0.00114
(-3.0869)***
-0.001265
(-3.0069)***
-0.00113
(-2.7355)**
-0.0011
(-2.39)**
-0.00115
(-2.619)**
3.02E-05
(0.0727)
-0.0003
(-0.502)
GDPR
GNPR
-0.00039
(-0.644)
-0.00068
(-0.9749)
RELGDPR
-0.0043
(-0.654)
-0.00352
(-0.8544)
YEAR
1.3875
(5.169)***
1.4697
(4.88)***
1.3796
(4.6488)***
1.369
(4.528)***
1.4344
(4.68)***
R2
0.6824
0.69
0.6825
0.69
0.703
DW
1.47
1.427
1.4744
1.41
1.3829
F
12.89
9.46
9.139
7.15
7.59
Prob F
0.000
0.000
0.000
0.000
0.000
Note: LFPR = Labor Force Participation Rate, TI = Transfers to Individuals, NTI = Net Transfers to Individuals,
GNPR = Real GNP, GDPR = Real GDP, RELGDPR = (GDPUSA/GDPPR)
* denotes significance at 10% level, ** denotes significance at 5% level, *** denotes significance at 1% level
24