Citation
Licensing of child care centers

Material Information

Title:
Licensing of child care centers
Creator:
Tinnin, Thomas Dean
Publication Date:
Language:
English
Physical Description:
v, 101 leaves : illustrations ; 29 cm

Thesis/Dissertation Information

Degree:
Master's ( Master of Science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Economics, CU Denver
Degree Disciplines:
Economics

Subjects

Subjects / Keywords:
Child care services -- Licenses ( lcsh )
Day care centers -- Licenses ( lcsh )
Child care services -- Licenses ( fast )
Day care centers -- Licenses ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 99-101).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Arts, Department of Economics.
Statement of Responsibility:
by Thomas Dean Tinnin.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
23281957 ( OCLC )
ocm23281957
Classification:
LD1190.L53 1990m .T56 ( lcc )

Full Text
I
LICENSING OF CHILD CARE CENTERS
by
Thomas Dean Tinnin
B.A., Metropolitan State College, 1985
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Arts
Department of Economics
1990
iJ


This thesis for the Master of Arts
degree by
Thomas Dean Tinnin
has been approved for the
Department of
Economics
by


Tinnin, Thomas Dean (M.A., Economics)
Licensing of Child Care Centers
Thesis directed by Professor Suzanne Helburn
The following paper addresses the effectiveness of
licensure as a means of correcting information asymmetry in
the child care market where product quality is thought to
be variable. Given the assumptions that quality in the
provision of child care can be identified and is measurable
and that licensure can indeed raise quality, the question
is then posed as to whether raising licensing standards
will benefit consumers, as indicated by their willingness
to pay for higher quality. An empirical test of the net
effects of raising licensing requirements is carried out,
following an approach introduced in the literature in 1987
for examining the role of professional interests in setting
licensing standards. The results are not encouraging for
those who believe raising licensing standards will benefit
consumers.
The form and content of this abstract are approved,
recommend its publication.
Signed
iii


CONTENTS
CHAPTER
1. INTRODUCTION ..................................... 1
A Note on Measuring Quality in Child Care ... 3
Thesis Outline.....................................5
2. LITERATURE REVIEW ................................ 8
Price vs. Quality Uncertainty .................... 8
The Role of Licensure.............................14
Quantity/Quality as Complements in
Production...................................16
Quantity/Quality as Substitutes in
Production...................................19
An Empirical Test of the Potential Benefits
of Licensure......................................29
Assumptions Pertaining to the Child
Care Market...................................... 32
Summary of Chapter.............................3 3
3. MEASURING THE COSTS AND BENEFITS OF LICENSING
REQUIREMENTS IN THE CHILD CARE MARKET . . . .35
Current State of Licensing Requirements
in the Child Care Market.......................3 5
Theoretical Model ............................... 38
Econometric Model ............................... 43
Dependent Variable
44


Independent Variables ........................ 46
Summary of Chapter................................61
4. RESULTS AND DISCUSSION............................62
Nonpolicy Variables ............................. 63
Policy Variables ................................ 66
Child Care as a Public Good . . .' . . 68
Licensure as an Means of Ensuring
Quality Provision .......................... 72
Summary of Chapter................................78
5. CONCLUSION........................................80
APPENDIXES
A. LELAND'S FORMAL MODEL OF THE EFFECTS OF
INFORMATION ASYMMETRY ON THE PROVISION OF
QUALITY...........................................85
B. THE COVARIANCE/CORRELATION MATRIX.................87
C. REGRESSION RESULTS WITH INCOME AS.DEPENDENT
VARIABLE..........................................89
D. REGRESSION RESULTS WITH X3 OMITTED................90
E. DATA TABLE FOR REGRESSION ANALYSIS................91
NOTES................................................94
SELECTED BIBLIOGRAPHY ............................... 99
v


CHAPTER 1
INTRODUCTION
Statistics on the growing number of women in the labor
force are widely cited. In 1987, the Department of Labor1
reported that 45% of all workers were women, up from 30% in
1950. In that same year, more than 70% of women aged 25 to
34 were in the labor force, whereas in 1950, only 35% were.
More significantly, in 1987, 57% of women with children
under age six worked, of these 70% worked full time. By
comparison, in 1950 only 12% of women with children under
six were in the labor force.
Overall participation of women in the labor force is
expected to increase, so that by the turn of the century,
61.5% of women will be at work. Three out of every five
new entrants between 1986 and the year 2000 will be women.
These numbers underscore the ever increasing need for
care of young children outside the home. This need comes
largely from two different groups of households.
The first group is composed of 12.8 million married
couples with both parents in the labor force, who have 8.8
million children under six. Three quarters of those
belonging to the group earn over $25,000 per year, and


receive little or no public assistance.
The second group consists of 3.5 million single .
mothers in the labor force, with 1.8 million children under
six. About 85% have incomes less than $25,000 per year,
60% with annual incomes below $15,000. About 16% receive
some public assistance. Of those on welfare, most do not
work and 1.5 million are exempted from work requirements
due to the presence of children under six.
What mode of child care do these households employ?
In 1985, 48% had their children cared for by
relatives, down from 62% in 1965. Only 6% had their
children cared for by a sitter in the home, down from 15%
in 1965. About 16% used family day care homes i'n 1965,
increasing to 23% in 1977, then declining slightly to 22%
in 1985. Finally, showing the largest increase by far was
center care, up from 6% in 1965 to 23% by 1985, with an 8%
increase between 1982 and 1985 alone.
Clearly there is an emerging market for child care
that is expected to continue to expand and to remain with
us in the future. However, markets do not simply arise in
a vacuum. In fact, for stable efficient markets to arise
at all requires a complex set of laws specifying property
rights along with the ability to enforce such rights
consistently. The specification of these rights is at
times hampered by the lack of information and the
2


possibility of nonexclusive attributes. These market
failures require intervention, but such intervention also
invites renewed efforts from special interest groups to use
the intervention to increase their own welfare at the
expense of the general welfare. Distinguishing beneficial
from detrimental intervention is a difficult task.
The child care market presents itself as one of the
areas challenging the analytical tools of the economist,
for the literature here is replete with the problem of
nonexclusive attributes, quality uncertainty, and asymmetry
of information between buyer and seller in the provision of
child care by nonrelatives. This paper addresses the
effectiveness of licensure as a means of correcting
information asymmetry in the child care market when quality
of service is assumed to be variable.
A Note on Measuring Quality in Child Care
It should be emphasized here that the primary purpose
of this paper is not in defining what quality child care is
nor in evaluating the appropriateness of those conditions
adopted as measures of that quality, such as child/staff
ratios, group size, and preservice educational requirements
for child care providers. Previous studies, notably the
National Day Care Study2 conducted in 1977-78 and the more
recent National Child Care Staffing Study3, have focused on
3


this issue of what quality is and how it can be measured.
For the purposes of this paper it is not necessary to
examine in depth the results of these studies, for this in
itself would require another paper. Rather, it is assumed
that there is sufficient evidence for a causal relation
between child/staff ratios or educational requirements for
child care providers and a child's behavior to warrant the
assumption that if these two measures of quality (i.e.,
child/staff ratios and preservice educational requirements)
constitute variations in licensing requirements then
stricter requirements may raise quality in licensed
establishments but in no circumstance would diminish
quality so long as these establishments are also in
compliance with the standards. The primary question of
this paper is whether the added costs (whether passed on to
the consumer in the form of higher prices, reduced quantity
or some combination thereof) of these higher licensing
requirements exceed the amount parents are willing to pay
for the resulting increase in quality that they would
expect to find in any given establishment covered by the
new standards.
Note that there is no presumption that the consumer is
certain the establishment is in compliance with the
existing standards at the time of purchase of services.
4


Note also that this wording does not specify a production
function for quality, wherein specific child/staff ratios
or years of preservice education, for example, result in a
given increase in quality, for which the marginal resource
cost could then be calculated. Rather, the intent of the
wording is simply to suggest a comparison the consumer
makes between the added value of the increase in per unit
quality (say, quality per hour) they would expect to find
in an establishment covered by the new licensing
requirements with the added value that the same dollar
would buy in the absence of the new licensing standards.
The distinction is important because it draws out the
nature of the uncertainty the consumer faces in evaluating
not only the quality of the service but the value of the
licensing standards. This point will become clearer once
the analysis proceeds.
Thesis Outline
In order to determine whether licensure is an optimal
solution to information asymmetry in the child care market,
the problem of information asymmetry must first be set out
in the paper. Chapter 2 will review some of the recent
literature on imperfect information and licensure as a
means of addressing this problem. The distinction between
price and quality uncertainty is illustrated first,
5


followed by a discussion concerning quantity and quality as
substitutes or complements in production and the difference
it makes for the rationale behind licensing requirements.
Certain assumptions are then adopted for the child care
market.
Chapter 3 will begin with an informal discussion
pertaining to the disparity in existing licensing
requirements observed across states. Several explanations
for this disparity will be offered, including confusion
regarding the effects of licensing in different industries.
Subsequently, both the theoretical and empirical models
used for ascertaining the potential benefits of licensing
requirements will be presented. Specifically, the model of
licensure represented by equation (2.5) in Chapter 2 will
be adopted for the child care market.
The explanation of the empirical model actually used
to test the net benefits of licensure in the child care
market will take up considerable space in this chapter.
Moreover, the rationale for predicting many of the effects
of the independent variables on the dependent variable will
be based upon the model of the household as a producer
unit, with own-time and market resources as the inputs into
certain basic commodities produced by the household. For
an overview of this model see, for example, Becker4 or
Schultz5.
6


Chapter 4 will present the results, followed by a
discussion. This discussion will revolve around two
possible explanations for the observed results, one
focusing on the divergence between the individual's and
society's willingness to pay for quality in child care, and
the second focusing upon licensure as an efficient or
least-cost method of raising quality in the child care
market. Both explanations will utilize the model of
licensure presented in Chapter 2.
Finally, the conclusion follows in Chapter 5 as an
extended discussion of the results obtained in the
preceding chapter, with suggestions for alternative means
of raising quality in the child care market rather than
through higher licensing standards.
7


CHAPTER 2
LITERATURE REVIEW
To understand the problems inherent in the child care
market a brief review of the literature on information
asymmetry is required. Most of the articles cited below do
not deal with the child care market itself, but rather with
a more abstract analysis of information asymmetry.
Assumptions and conclusions regarding this background
literature will be brought to bear on the child care market
later in this chapter.
Price vs. Quality Uncertainty
In 1961, Stigler published "The Economics of
Information"6 in which he hypothesized a tradeoff between
the price of the product and the cost of continued searches
for lower prices, resulting in the possibility
of equilibrium price and quantity deviating from
competitive pricing models. Given that a consumer knows
the average price in the market and measures of the
dispersion of prices, then searches for prices below the
average price will continue until the expected reduction in
the price from the last search times the quantity purchased


is just equal to the interest rate times the marginal cost
of search, or
(2.1)
where the marginal cost of search consists largely of the
value of the consumer's time. Clearly a certain amount of
price dispersion may persist in equilibrium depending upon
the costs associated with searches in a given market.
An entire literature grew up around this analysis.
For example, subsequent models included the behavior of the
firm under price uncertainty. Models incorporating price
and wage uncertainty have now become integral to theories
dealing with frictional and cyclical unemployment.
Although price uncertainty may or may not exist in the
child care market, it is mentioned here as an important
prelude to the notion of quality uncertainty in markets
where certain nonpriced attributes of a commodity vary from
one unit to the next, and where this attribute is valued by
consumers. For example, orange juice may be sold by the
quart, gallon, etc., and although each quart may be priced
the same they may differ slightly in terms of such
nonpriced attributes as sweetness, pulp, freshness, and so
on. In fact, the more complex the product or service, the
greater the chance of quality variation among units.
9


The primary difference between price and quality
uncertainty is that quality uncertainty is ordinarily
detected only after consuming the good or service, .and
frequently there is uncertainty even then, as for example,
with physician services or medicinal products. This makes
a difference in search strategy,, for it makes little sense
to conduct prepurchase searches if the relative quality of
the products cannot be readily compared. However, other
costs do arise for quality uncertainty, such as monitoring
the service, switching to new suppliers, specifying
contracts, or seeking recompense in the court system for
substandard goods or services.
The article most often cited as dealing explicitly
with quality uncertainty is Akerlof's "Market for Lemons"
article published in 1970.7 He illustrates how newly
emerging marketsparticularly in developing countriesmay
be severely hampered by an asymmetry of information that
develops between buyer and seller regarding the nonpriced
attributes in any particular commodity offered for sale.
Since these attributes can be known only subsequent to the
purchase the owner of such a commodity is presumed to know
more accurately the true value of the commodity. Moreover,
because the consumer cannot trust the seller to reveal this
value (a case of moral hazard), market transactions may be
severely limited or even nonexistent even though there are
10


potential gains from such trade.
Akerlof then sites the emergence of product
guarantees, brand names, service chains and licensing as
(presumably the most unobtrusive) institutions arising to
facilitate exchange.
Clearly, either quality or price uncertainty can exist
without information asymmetry, though this is not usually
the case. For example, it may be the case that neither
child care consumers nor providers know what constitutes
quality care, particularly in an environment outside the
home. However, the assumption adopted here is that the
seller has a more accurate indication of the relative
quality of the services he provides than the consumer does.
To better understand the distinction between price and
quality uncertainty, Figure 2.1 is reproduced from the work
of Krashinsky.8 The line OB represents the production
possibility frontier, or the supplier's marginal
opportunity cost of providing a given quantity of a good or
service or increasing quality per unit. For example, qL
might equal one hour of a low quality service, say a
tutorial, while qh might represent the same hour but of a
higher quality. The line OB is the locus of least cost
combinations of inputs into quality provision (here
reflecting increasing marginal cost of quality provision)
in a competitive market absent of either price or quality
11


uncertainty. In such a Price,
market this line will Cost
represent both the ph T c / i
marginal cost and the 1 / i / i i i
selling price of units of n i/ i
pi i
consecutively higher j / 1 i
quality. However, when 0 j 1 l 1
uncertainty enters the ^1 %ukll
picture, price and Figure 2.1. distinction A between
marginal cost will diverge. In the case of pure price and uncertainty. quality
price uncertainty the seller produces at point A but sells
at point C, i.e., the consumer is aware of the relative
quality of the good purchased but not aware that a greater
quantity (say, two hours) is available elsewhere for the
same price. Note that since production is still on the
line OB there is no inefficiency in production. Whether
the price differential Ph Pt diminishes or disappears
depends upon the size of the market, the relative stability
of supply and demand, characteristics of the product, and
the fraction of income devoted to purchasing the product,
to mention the most important determinants.9
Under quality uncertainty, however, the seller
produces at point C a good of quality qt and sells it a
12


price Ph. The consumer knows of the existence of price Pt
for the same quantity (i.e., for a one hour of tutorial)
but is fooled into believing he is purchasing a higher
quality service. The inefficiency of producing at point C
stems from the production of specious market signals (S) so
long as Ps < Ph PL. The investment in market signals
therefore represents a case of rent-seeking behavior. Low
quality providers are induced to purchase these signals so
long as higher quality producers have an incentive to
distinguish their product from low quality sellers.
Moreover, Spence10 illustrates that for these signals to
effectively discriminate between high and low quality
sellers, the cost of acquiring the signal must be lower for
high quality producers. (In fact, the signal need not
enhance quality for it to be an optimal solution to
imperfect information.)
In the case of quality uncertainty the price Ph will
just support the marginal seller of quality qh as it is
equal to his marginal opportunity cost. But it will also
support all intramarginal sellers of quality q < qh as long
as consumers are unable to distinguish quality at the point
of purchase and there is little or no accumulation of
knowledge in the market that serves to differentiate low
and high quality providers. (This is an unrealistic
assumption but serves to make the point.) Since consumers
13


cannot distinguish high from low quality but are aware that
such a range exists in the market, then an equilibrium
price such as Ph must reflect the consumer's marginal
valuation of the per unit quality they expect to find in
the market, i.e., the average quality. Yet because the
marginal seller provides a unit of above average quality
the marginal social benefit of his service exceeds his
opportunity cost, or Ph, and less than the optimal amount
of quality will be provided. Leland11 obtains this result
in a formalized model of Akerlof's 1970 outline, the former
here reproduced in Appendix A.12
The question then arises, what institutional
arrangements must evolve to ensure a more optimal provision
of quality at the lowest cost? Licensing has emerged as a
popular solution in many industries, including child care,
where all states have now adopted licensing standards for
child care centers.
The Role of Licensure
Licensing invokes the police power of the state to
protect consumers from fraud or negligence. The licensing
standards may be set and enforced either by members of the
profession or by an outside regulatory agency. In the case
of day care, the state ordinarily assumes the role of
setting and enforcing standards governing the operation of
14


child care facilities. As such, licensing takes on the
guise of a welfare service. As Class and Binder write:
The ultimate purpose of welfare licensing is
fundamentally no different from the ultimate
purpose of professional licensing for such groups
as doctors or pharmacists. In each instance the
community, speaking through the legislature, has
decided that the users of the services concerned
are not, by themselves, in a position to inquire
into and properly appraise the standards under
which the service is rendered.13
This view of licensure stands in contrast to the
alternative view that licensure is a form of rent-seeking.
However, pointing out the potential benefits of licensure
does not mean that rent-seeking does not exist. The
assumption adopted here is that there is always an
incentive for producer groups to obtain rents from the
provision of their services. What discourages most groups
from acting to obtain potential rents, of course, is the
relative costs of organizing, lobbying, and so on.
Presumably, legislators are encouraged to pass
licensing, laws on the basis of such arguments that
consumers as well as special interests may benefit from the
laws. However, in spite of the general intent for
licensing to increase general welfare, there appears to be
considerable differences in the literature on how licensing
actually affects the market and whether it is the best
solution to the market failures outlined earlier.
The effect licensing has on the behavior of market
15


participants may be different from the effect it is
intended to have. One of the problems stem from the fact
that a market characterized by quality uncertainty is
comparable in nature to collapsing two or more markets for
homogeneous goods into a single market where many of the
participants cannot distinguish variations in quality until
after purchase, nor perhaps have the opportunity of
learning to distinguish through repeated purchases.
Nonetheless, the value of these nonpriced attributes
reveals itself through its impact on quantity exchanged and
the market price, much like substitutes or complements
impact across markets for homogeneous products. However,
the graphical depiction of a multi-dimensional product in
a two-dimensional space is not quite as straightforward as
this analogy would suggest. This is due to the added
complications of representing informational asymmetry and
the exact relationship between quantity and quality.
Ouantitv/Oualitv as Complements in Production
If quality cannot be varied independently of quantity
but is instead an inherent attribute of the product or the
person providing the service, then quantity and quality can
be thought of as perfect complements in production. In
this case, increased per unit quality in the market can
only come about by increases in the number of units
supplied through entry of marginal sellers offering the
16


higher quality products. This is how quality variation is
modeled by Akerlof and Leland.
Since quality is a monotonic function of quantity
supplied, and the latter a function of price, then
variations in the average quality on the market can be
represented by an upward sloping supply schedule in the
price-cruantitv space, as depicted in Figure 2.2.
On the demand side, quantity demanded (Qd) is a
function of price (p) and average quality (q) in the
market, or
Qd=D (p, q(p) ) (2.1)
where
dD=dD+dDxdq(pl {23)
dp dp dq dp
The first term on the right is negative and the last
two terms are positive, which means the demand schedule may
be positively sloped if the second term is larger in
absolute value over some range of prices, as depicted in
the graph. Obviously the price is itself serving as a
market signal for expected quality, wherein higher prices
signal higher expected quality, and therefore higher
expected utility. However, as price continues to rise the
fall in the marginal value of the increased quality per
17


unit eventually offsets
the increases in per unit
quality and the demand
schedule reverts to its
normal negative slope.
Although there may be
multiple equilibria, the
one depicted here
represents the point
where expected marginal
utility of the last unit
equals the marginal cost
of the last unit.
Price
Figure 2.2. Market
with asymmetry of
information and
quantity/quality are
complements in
production.
However, as mentioned before, the quality offered by the
marginal seller is greater than the average quality in the
market, and therefore, marginal social benefits exceed the
supplier's marginal private benefit and there is under-
provision of quality at Q.
When quantity and quality are complements in
production, the primary problem is adverse selection of low
quality producers. Hence, the intent of licensing under
this situation is to screen out the lowest quality
providers, thereby raising average quality and price. This
may be represented as a shift upward of the supply curve
(as the lower end of the old supply curve is simply
18


truncated at the level of minimum standards). As is
apparent from the diagram, it is possible for quantity
supplied to actually increase.
This is perhaps an unexpected benefit of licensing
provisions that restrict entry and is due solely to
information asymmetry. Moreover, as long as quality cannot
be varied independently of quantity, then licensing is
likely to be superior to (seemingly) less obtrusive
measures such as product guarantees or contingent
contracting because licensing eliminates the poorest
quality at the source, rather than requiring consumers to
specify contracts and monitor quality each time they
purchase. Even if the screening of applicants approaches
random selection of high and low quality providers,
licensing will still raise average quality in the market
since the increase in price will attract higher quality
applicants.
Quantitv/Oualitv as Substitutes in Production
However, the case is more complicated if quality can
be varied independently of quantity supplied by existing
suppliers. In this situation, even if providers are
screened prior to entry into the market, there is no
guarantee they will maintain the quality standards set for
their product once in the market. Hence, the license must
serve to dissuade quality deterioration, and this could
19


call for frequent monitoring of the service by an
independent group. Whereas before quality deterioration is
ruled out once low quality providers are denied entry, here
the problem remains a possibility.
In this case, quality and quantity can be thought of
as imperfect substitutes in production. Graphically,
instead of depicting quality as varying along the supply
schedule, quality variation would here be represented by a
shift in the supply schedule, with each schedule defining
a given level of quality. Thus, if q = Q/N where Q is
total quality supplied and N the total quantity or number
of units supplied, then each supply curve represents
various quantities of a given quality, or q. If the
marginal cost of quality provision is increasing, then a
shift upward in the supply curve represents a new set of
products on the market with higher per unit quality, with
the amount of the shift being the marginal cost of q.
Likewise, marginal valuations of increments in quality
per unit would be represented by shifts in a negatively
sloped demand curve, wherein each schedule is defined as
demand for a set of products with per unit quality q
constant. An upward shift in the demand schedule
represents demand for a new set of products with a higher
per unit quality q, with the amount of the shift
representing the marginal valuation of the increase in q.
20


With quantity (N) and quality (Q = qN) the two
arguments in cost and utility functions the efficient
combination is found by maximizing the difference between
consumer's valuation and producer's costs, or
MaxNi0 u [N, qN) -c(N, qN) (2.4)
This is tantamount to maximizing the area between the
supply and demand curves, or the sum of producer's and
consumer's surplus, in the price-quantity space. Leffler14
adopts this approach but for his purposes does not assume
the persistence of any information asymmetry, so that after
repeat purchases the equilibrium in the price-quantity
space reflects a social optimum.15 For the purposes of
this paper information asymmetry is assumed to persist to
some degree so that at any point in time the position of
the demand curve is, in effect, lower than it should be,
reflecting the marginal utility of the average quality
rather than the marginal quality offered. It is apparent
then, that increased information will shift the demand
s chedu1e upward.
But here another complication arises. Normally
increased information can only be obtained at a costin
this case the time and resources of the consumer in
researching the market or perhaps in monitoring services.
Does the vertical distance between two demand curves
21


represent the marginal valuation of the increase in
expected quality alone or of the marginal valuation of the
increase in expected quality and the value of the time and
resources used in ascertaining the average quality in the
market and any changes therein?
Some economists, notably Akerlof and Leland,16 adopt
a rational expectations hypothesis in that consumers always
know (without cost) the average quality prevailing in the
market yet have no firm specific knowledge of quality,
i.e., they do not know the relative quality of this
particular product or service. If this is true then the
shift in demand curve may represent only the marginal
valuation of the change in expected or average quality, for
since the consumer always knows the expected quality (as
well as measures of dispersion of quality) and prepurchase
searches are futile because the relative quality cannot be
judged, then only one contact is called fordrawn randomly
from the available supplierswhich remains as costless as
before the change in market quality.
While this assumption is convenient for the analyst,
it could hardly be an accurate description of consumer
behavior, especially when quality uncertainty exists. To
know the expected quality consumers must at least know the
relative frequency of each level of quality available in
the market, which must then be further subdivided into
22


local markets, regional markets, and so on, each perhaps
differing in average quality. Consumer reports ordinarily
focus on national or regional markets, which may be of
limited help to those utilizing local markets.
Additionally, even if prepurchase searches in the case of
quality uncertainty are ruled out, there are still other
costs associated with securing higher than average quality
in the market, as e.g., monitoring services or specifying
contracts. Because of these costs, consumers must know
measures of dispersion in order to estimate the net
benefits associated with monitoring a service or specifying
contracts. For example, if the chances of selecting a poor
quality establishment are virtually negligible then
extensive monitoring and specification of contracts may not
be warranted.
Secondly, even if the consumer always knows the
average quality and degree of variance in the market,
selection may nonetheless be purposive rather than random,
reflecting the existence to some degree of established
reputations, referrals, location, and so on. In this case
we may still speak in terms of an expected quality, but
only in regard to each consumer's subjective expectations
and not in terms of any one "objective" value for the
market held by all consumers.17
In short, information of any type, especially that
23


concerning average quality, is not costless to obtain, and
any market signal that lowers the cost of market research
and monitoring has value to the consumer. If this is the
case then any shift in the demand curve that reflects the
marginal valuation of the higher expected quality may also
represent the value of any savings in transactions costs
associated with ascertaining the (new) expected or average
quality in the market and in locating a particular provider
of average or higher quality.
Shifting attention now to the supply side, it is
apparent that the problem is not so much one of adverse
selection of poor quality providers as it is regarding the
adverse incentives of existing providers to lower quality.
That providers, given a choice of varying quantity or
quality to reduce costs, would choose to lower quality is
apparent, since such variations are difficult for the
consumer to perceive. But if one lowers quality, all have
an incentive to do so, and all are hurt by the lower price
that each may ultimately receive, resulting in lower
quantity as well. Raising quality suffers from the same
sort of dilemma. Each provider bears the full cost of
higher quality but allincluding the provider who does
raise his qualitywill benefit from the higher average
quality and the higher prices received. However, when the
number of providers is large, the increase in average
24


quality from one provider raising quality approaches zero,
and there is no incentive to raise quality voluntarily.18
To summarize, when quantity and quality are
substitutes in production it becomes apparent that
licensure must accomplish more than merely screening out
those providers unable to supply a minimum level of
quality. Now quality variation is a decision variable.
Licensure must somehow adjust the incentives facing
providers so that they will produce and maintain a minimum
level of quality.
One of these incentives is the possibility that if
licensure calls for a minimum of preservice education or
training the marginal costs of providing higher quality
will decline with increases in human capital. Thus,
Shapiro19 sees licensing as an input regulationtrained
providers will prefer to supply higher quality goods or
services because it is simply more profitable than
providing low quality.
Although decreasing marginal costs for quality may be
a reasonable assumption for some markets, it is unlikely to
be the case in all or even many of the markets where
asymmetry of information exists.
If decreasing marginal costs of quality provision are
ruled out for a particular industry, as Blair and Kaserman
do for example when stating that "the payoff for any given
25


quality deterioration must take the form of reduced cost of
operation and/or increased revenues,1,20 then the rationale
for licensing shifts to the purported "wealth" effect for
the provider. Again, Blair and Kaserman write "...society
pays a bribe in the form of noncompetitive fees [rent from
the license] and then hopes that this will prevent low-
quality service."21 They construct an incentive function
I = (W, w, p, S) to depict the behavior of a supplier when
faced with the opportunity to lower quality in order to
increase profits when there is the risk of a sanction.
This incentive function takes the form
I=E[ (1 -p) U(P/+w) +pU(W+w-S) ] -U(W) (2.5)
where
I = incentive to reduce quality
E = expected utility of wealth
U. = von Neumann-Morgenstern utility function showing risk
aversion
W = "legitimate" wealth from preserving quality standards
w = added wealth from quality deterioration
S = magnitude of sanctions imposed for quality
deterioration
p = probability of being detected for quality
deterioration (0 < p < 1)
Clearly, if I(W, w, p, S) > 0 (e.g., when S < w
regardless of the value of p or when S > w and p is
sufficiently small) there is a positive incentive for
quality deterioration. The primary effect of licensing is
in raising legitimate wealth (W), with the magnitude of W
26


rising with increases in the licensing standard. Blair and
Kaserman show that even though licensing may not eliminate
quality deterioration it may reduce it if
(2.6)
which may or may not be the case depending upon the
relative magnitude of the sanctions (S) and the probability
of being detected (p). In fact, the authors conclude:
If one wants to rely on increasing wealth to
bribe professionals to refrain from reducing
quality, the best thing to do is to have only
light penalties (small values of S) and minimal
enforcement (low values of p) ,22
This unexpected result can be seen from examining the
partial of I with respect to W, or
= [U'(W+w) -U'(W) ] -p[U'(W+w) -U'(W+w-S) ] (2.7)
The first term on the right is negative due to the
assumption of risk aversion. However, with S > w and p
increasing the sign of the second term becomes negative,
which is then subtracted from the first creating a positive
influence that may outweigh the negative influence of the
first term. As the authors explain, this result simply
reflects the tendency for wealthier individuals to accept
greater risks.
27


The above result, however, does not change the
expected results that , given any level of W,
aw |I<0 Bp (2.8)
that is, for any given minimum standard (associated with a
given amount of legitimate wealth) increasing the amount of
sanction or the probability of detection reduces the
incentive to lower quality below standards.
Only licensing will create the wealth effect suggested
above, for it excludes those not holding a license from
providing the service and erects a potentially effective
long-term barrier to entry that preserves the future stream
of above normal net returns. Similar market interventions,
notably certification, cannot accomplish this, because
certification does not entail exclusion of those programs
where the degree of quality has not been ascertained by an
independent authority.23 Hence, the wealth effect
associated with licensing is a reply to arguments asserting
that certification is a more efficient means of correcting
quality uncertainty for the very reasons that it does not
restrict entry and provides the consumer with a greater
range of choice.24
It would appear, therefore, that when quality can be
varied independently of quantity and the marginal resource
28


costs of quality provision are increasing, the weight of
the argument in favor of licensing would rest upon this
wealth effect.
An Empirical Test of the Potential Benefits of Licensure
Now that the theoretical benefits of licensure have
been outlined, what has been the empirical evidence cited
in the literature for licensure?
Of particular interest for this paper is an article
published in 1987 by Shirley Svorney25 addressing the
impact of licensure on the market for physician services.
Svorney wished to devise a test of whether licensing
standards set for physicians did indeed result in net
benefits to the consumer, as opposed to benefiting only the
professional interests. Svorney reasoned that licensure
was superior to certification primarily along the lines of
the preceding discussion, i.e., licensing creates a wealth
effect that reduces the incentive to lower quality. Hence,
Svorney is implicitly assuming that quality is a decision
variable for physicians currently practicing in the market
and that the marginal costs of quality provision are
increasing. These assumptions appear to be warranted,
considering the nature of the service.26
However, Svorney emphasized the threat of losing a
license rather' than the positive effect the above normal
29


returns has on the provider. The difference in perspective
may seem inconsequential, but with the emphasis on
sanctions questions regarding the costs of enforcing the
sanctions arise. Thus, for a minimal or modest licensing
standard, an increase in sanctions (S) or the probability
of being detected (p) will reduce the incentive to lower
quality below the standard (though not for those with
quality initially above the standard) but these measures
will also entail increased costs in monitoring and
enforcing the standard.
On the other hand, Blair and Kasermen seem to be
saying that if the licensing standard is set high enough
one need not worry about the effectiveness and hence the
cost of enforcing sanctions, assuming that higher standards
are positively correlated with higher rents for existing
providers. But once again, there are definite costs
involved for setting standards high, especially for
consumers who prefer low quality care to no market care.
In either case, the costs of the licensing must not
exceed the benefit the license entails for consumers. This
is tantamount to saying that the (exogenous) shift upward
in the supply curve due to licensing restrictions should
not be greater than the (endogenous) shift upward in the
demand curve. As changes in demand as a function of the
severity of licensing restrictions can ostensibly be
30


observed, then the net benefits of regulations restricting
entry into markets with asymmetry of information can in
principle be determined by observing the direction of the
change in equilibrium quantity exchanged. Thus, as Svorney
states, even though licensing will restrict entry and
raise prices,
as long as the increase in the supply price of
physician services caused by medical training
requirements is less than the value of the added
assurance of performance, medical entrance
requirements will benefit both consumers and
physicians and the total quantity of physician
services consumed will increase.2'
Clearly, with the phrase "value of the added
assurance," Svorney is including the value of the time and
resources saved in gathering information on quality in the
market. The rational expectations hypothesis is not being
invoked. Consequently, any shift in the demand curve would
include this value. On the basis of the preceding
discussion on markets where the marginal cost of quality
provision is increasing, this method would appear to be
valid.28
The object now is to further define this empirical
test and to apply it to the market for licensed day care
centers.
31


Assumptions Pertaining to the Child Care Market
If Svorney's empirical model is to be applied to the
child care market, this in turn implies that certain
assumptions regarding the market for child care are being
adopted, notably, that
1) Quantity and quality are substitutes in the
of child services, or that quality is a decision
variable, and
2) The marginal resource costs of quality provision
are increasing in the child care market..
The first assumption appears to be warranted by casual
observation. The second assumption appears to be warranted
by the equally casual observation that cost reducing
measures such as increasing the child/staff ratio, hiring
low paid, low skilled teacher's aides, or increasing group
size are normally associated with lower quality care. Of
course, these assumptions rest upon a still more basic
assumption that quality can indeed be identified and
measured in a complex service such as child care.
Therefore, before any empirical tests are reported, the
next chapter will begin by examining the rationale behind
current licensing requirements in the child care market.
Once this rationale is outlined, it can then be tested.
32


Summary of Chapter
In this chapter the nature of the effects of quality
uncertainty and information asymmetry on market prices and
quantity exchanged was explored. Specifically, when the
seller knows more about the relative quality of the good or
service being sold, then market prices and quantity
exchanged both tend to be lower than optimal, and in
extreme cases, no trade at all will occur.
Moreover, in markets where information asymmetry
exists, there is a marked tendency for lower than optimal
provision of quality per unit of good or service.
Specifically, when quantity and quality are complements in
production, i.e., quality is an inherent or fixed attribute
of each unit of good or service sold on the market, only
the lowest quality goods and services will be sold. In
other words the market adversely selects the lowest
quality. On the other hand, in markets where quality is
primarily a decision variable, i.e., can be varied
independently of quantity sold, there will still be less
than an optimal amount of quality provided in the market
because of the public good attributes quality provision
entails. The distinction bears upon the intended effect
licensure is presumed to have for either type of market
suffering from information asymmetry.
33


In the case where quantity and quality are complements
in production, licensure is intended primarily as a
mechanism to screen out lower quality providers, thus
raising average quality, price, and presumably demand,
since the latter is a positive function of average quality.
In the case where quantity and quality are substitutes
in production, it was determined that the primary effect of
licensure was entailed in a "wealth" effect that served to
discourage existing providers from lowering the quality of
their product on an public unable to adequately discern
variations in quality at the point of purchase. Equation
(2.5) represents this model of licensure.
Thereafter, it was assumed that for the child care
market quality could be varied independently of quantity
and that the marginal resource cost of quality provision
was increasing. Hence, the intended effect of licensure in
the child care market is represented by equation (2.5).
Fortunately, the effects implied by equation (2.5) can
be represented graphically on a set of supply and demand
curves, which is exhibited in Chapter 3. Moreover, the
results obtained graphically can in principle be tested for
in the real world, and Chapter 3 specifies this empirical
model.
34


CHAPTER 3
MEASURING THE COSTS AND BENEFITS OF LICENSING
REQUIREMENTS IN THE CHILD CARE MARKET
This chapter begins with a brief and informal overview
of the existing disparity in licensing requirements across
states. This initial discussion is intended to provide a
background to prepare the reader for the theoretical and
empirical models developed in the remainder of this
chapter. These models are used to obtain estimates of the
net benefits of licensing requirements.
The Current State of Licensing Requirements
in the Child Care Market
Regulation of child care markets varies considerably
both within and among states. Currently, many states only
license centers and not family day care homes (FDCHs), even
though the two may be considered substitutes. Moreover,
even though all states do require centers to be licensed,
the licensing standards vary considerably from one state to
another. This situation does not appear to reflect an
understanding of what licensure is or what it is supposed
to accomplish.
Focusing on the variation in licensing standards


reveals several different standards corresponding to
several different levels of quality. Minimal standards can
be thought of as pertaining only to site characteristics
fire safety, sanitation, etc. Higher standards may then
begin to place limitations on child/staff ratios and the
group size, with variations in these restrictions again
intended to reflect different degrees of quality. Still
higher standards will ordinarily include preservice
requirements for directors and teachers, ranging from a few
hours of orientation to college degrees. Although costs
associated with meeting minimal requirements are not
insignificant, the preservice educational requirements can
entail substantial investments, especially if college is
involved.
Do consumer tastes and production technologies for
quality child care really vary across states to the extent
implied by the variation in licensing standards?
One alternative explanation cites the role members of
the licensed industry have in setting standards. For
example, Moore29 conducted research showing that standards
set by members of the profession tend to be set higher than
those set by an outside regulatory agency, such as the
state or county. Standards for child care licensing,
however, are usually set by the state.
If standards are set by public officials, then a
36


different set of explanations for the existing disparity in
licensing standards can be invoked. The first draws upon
the rent-seeking view of licensure, wherein variation in
lobbying strength for those whose interests may be served
by stricter licensing requirements accounts for the
variation in licensing standards. But then variation in
lobbying strength across states must be explained, and the
explanations here can be numerous. For example, it may be
the case that even though smaller organized groups with
considerable resources can bring more pressure to bear on
public officials than consumers or voters in general,
elected officials nonetheless also have an incentive to
respond to their constituent's interests, especially when
the demands of special interests begin to incur significant
deadweight costs on society.30 But suppose public
resistance to the demands of existing child care providers
to raise wages or limit new entrants is lessened by
assertions that stricter licensing requirements entail
significant benefits for consumers as well as producers.
Indeed, the implicit assumption running throughout this
paper is that the care of children should not be left to an
industry driven by profit when quality uncertainty exits,
for it is precisely this concern that licensing of for-
profit centers is supposed to assuage.
Perhaps a better explanation for the variation in
37


standards set by legislators is the confusion over what
licensure is supposed to accomplish or how it will
accomplish the desired outcomes. For instance, licensure
may be perceived as a screening device for new entrants,
with little need for constant monitoring once the applicant
is granted admission. If quality remains a decision
variable even after entry, then the totals costs associated
with setting licensing standards may have been
underestimated. On the other hand, the need for extensive
monitoring may have been recognized; subsequently, minimum
requirements may be intentionally set high in some states
to take advantage of the purported wealth effect outlined
in the previous chapter, while other states may have chosen
to keep standards low and allocate the bulk of resources to
enforcement.
Whatever the reasons for the disparity in licensing
standards, this paper presents models designed not only to
clarify the effects licensure is presumed to have but also
to test whether the' net effects are in the interests of the
consumer. The remainder of this chapter is devoted to
developing these models.
Theoretical Model
Now that the current state of licensing requirements
across states and the possible reasons for variations in
38


these requirements have been discussed, a test similar to
Svorney's can be constructed to determine whether more
stringent requirements observed in some states actually
benefit consumers.
Given an increase in a state's educational
requirements for directors, a decrease in supply of
licensed center care would be expected, here represented by
a shift of the supply curve from S to S1 in the Figure 3.1
below. Likewise, a lower child/staff ratio should shift
the supply curve upward as well. What effect will these
requirements have on consumer welfare as reflected in
changes in consumer demand?
There are three possible impacts on consumer demand as
a result of the stricter licensing requirements, depending
upon the relative amount of value such requirements provide
consumers.
The first possible impact is that the demand curve
shifts outward to, say, D'. However, the added per unit
value to the consumer is P2 PQ, which is less than the
added per unit cost incurred by the provider in meeting the
requirements, or the amount P1 PQ, thereby forcing some
providers out of the market. Note that total receipts may
(or may not) have risen, but quantity has declined, from Q0
to Qe. With a higher price and lower equilibrium quantity,
the only ones to benefit from such a provision are those
39


requirements on supply and
demand schedules.
providers remaining
in the market.
Moreover, there has
been a loss of
consumer and
producer surplus
(the relative amount
borne by provider
and consumer
dependent upon the
relative
elasticities of
supply and demand)
equal to the area
ABCD, while the total gain in producer and consumer surplus
is only equivalent to the area DEFG. Everyone, except
those cited above, is worse offnot necessarily by the
licensing per sebut by the excessive requirements imposed
by the licensing authority. Of course, it is possible for
the demand curve to not shift at all, meaning that the
requirements entail no value for the consumer. In a
regression, either possibility is represented by a negative
relationship between equilibrium quantity and a measure of
preservice licensing requirements for directors.
The second possibility is that the demand curve shifts
40


outward to D", in which case the added per unit value to
I
the consumer, or P, Pn, exceeds the added costs to the
I
provider, or P1 PQ. In this case, both consumer and
producer benefit as the difference in per unit price
received by the supplier is greater than the cost per unit
incurred, and the added per unit price paid by the consumer
is less than the added value derived from the requirements.
The combined gain in consumer and producer surplus is equal
to the area DEHI. Note that quantity exchanged has
increased, from Q0 to Qe 1,commensurate with the notion that
trust and improved information facilitate the expansion of
a market. In this case, a positive correlation between
equilibrium quantity and the measure of preservice
requirements would be expected.
There is, of course, a third possibility, namely, that
there is no significant correlation between equilibrium
quantity and preservice requirements. This means that
either the requirements affect neither supply nor demand
which is unlikely if the requirements are to serve any
purpose at allor that the increase in demand just offsets
the decrease in supply, resulting in a higher equilibrium
price and about the same quantity provided. In this case,
a second regression testing for a significant positive
correlation between licensing requirements and gross
revenue would serve to confirm the latter interpretation.
41


Given that the value of the requirements to consumers
is equal to the additional costs, consumers as a whole
would appear to be as well off as before the stricter
preservice requirements. The implicit costs associated
with child care have simply been incorporated into the
market price to reflect the actual costs borne by
consumers. However, this does not mean that each
individual consumer is equally affected by the new
requirements. The differential effects of stricter
licensing in this case stem from differences in each
consumer's production function for information and their
tastes for quality. If searching for and monitoring child
care quality are time intensive activities, then those
households where own-time was less expensive relative to
market goods will now be worse off, because it was cheaper
for these households to use their own-time in searching for
and monitoring quality rather than to purchase it as part
of the price of standardized licensed care, which is what
was entailed in the preservice requirements. Also, it is
apparent that those consumers with a higher marginal
valuation of a given quality will benefit from the higher
standards, while those with lower marginal valuations of
the same amount of quality will lose from the higher
standards, because they now cannot purchase quality below
the legal standard. (It is this diminution in choice for
42


these consumers that isj often cited as an argument in favor
of certification.)
Consequently, it is likely that the composition of the
consumers of center care would have changed given no other
changes in money incomes or child care subsidies during the
same period of time. This means that lower income
households may have been replaced by wealthier households
in the regulated market (especially if tastes for quality
are positively correlated with income) creating the
prospect of a dual system of child care, consisting of one
market that is essentially unregulated but with lower
average prices and used by those households who use their
own-time in searching for and monitoring child care
quality, and a second market with higher prices and
standardized quality used by those whose own-time is
relatively expensive and opt to purchase quality rather
than search for it or monitor the providers. This effect
may also be present to varying degrees in the other two
cases as well, since in all three cases a higher
equilibrium price is expected. More will be said on this
in the conclusion.
Econometric Model
To test for these outcomes, an ordinary least squares
regression was run with a dependent variable of per capita
43


licensed center slots within each state and ten independent
variables. The econometric model is
yi=a+hZil+cZi2+dXi2+eXu+fXl5+gXi
+hXil-kXiB+lXi9+mXil0+ei i = obs (3.1)
where Y is the dependent variable and Xj the independent
variables, all defined below. The sample range included
observations across 50 states and the District of Columbia,
excluding those states with missing values in the data set.
Dependent Variable
The data for the dependent variable was taken from the
1987 State Child Care Fact Book31 by the Children's Defense
Fund. Licensed center slots was chosen because it is the
most comprehensive measure of the quantity of child care,
as opposed to number of centers. The latter measure would
not pick up variations in size of existing centers or the
adjustments existing centers could make in the face of a
change in licensing requirements for directors.
Furthermore, the data on total licensed slots was
transformed to per capita to avoid the likelihood of
heteroskedasticity in comparing total' licensed slots
between states such as Rhode Island and California. Family
day care homes (FDCHs) and group homes were excluded from
44


the study because the majority (some estimates are between
50% and 75%)32 are not licensed or difficult to monitor if
licensed, and because the numbers of slots may not be as
reliable as those for centers, due to the fact that FDCHs
vary much more in capacity utilization than centerswhich
frequently operate at or near licensed capacityand
because entry and exit for FDCHs is much easier, rendering
records of licensing authorities subject to more rapid
obsolescence.
There were, unfortunately, disadvantages to using the
CDF data. The CDF data does not distinguish between
nonprofit and profit centers, nor between centers
participating in federal and state funding of child care.
While both profit and nonprofit centers are equally bound
to uphold licensing requirements, it can be argued that
for-profit centers are more relevant to the hypothesis here
invoked. Nonprofit businesses by definition are those
willing or able to forego a normal rate of return on
business investments, which would include educational
requirements. This suggests that nonpecuniary returns are
an important incentive for providing and maintaining the
service for others. Save for the possibility that these
nonpecuniary benefits are not in the interests of others,
this implies that such a provider may be willing to make
substantial investments to meet licensing requirements
45


without being sensitive to its internal rate of return, and
so continue to offer the service to consumers who are
unwilling to support the licensing requirements through
higher prices. Our hypothesis assumes this sensitivity to
the demands of the consumers through their willingness to
pay, and is therefore more appropriate to for-profit
businesses.
In spite of these considerations, however, comparison
with the 1987 U.S. Census of Service Industries reveals
that roughly two thirds of all child care establishments
with payrolls in the U.S. were subject to taxation,
indicating they were for-profit businesses. Under the
assumption that all centers have payrolls the CDF data
should provide a fair indication of the effects of
licensing restrictions on for-profit centers.
Independent Variables
For independent variables, the following constituted
the initial set formulated by theory. All variables are
across states.
Per capita disposable income.33 The inclusion of
some measure of income across states is obvious, but the
effects of income on the composite demand for child
servicesdefined as number of children per household times
the expenditure (both nonmarket time and market goods) per
childare often complex and hard to predict.
46


Census data over the past few decades have revealed
either a negative correlation between per capita income and
fertility rates, or a nonlinear relationship, wherein
fertility rates initially decline then begin to rise as a
function of increasing income. These data have led some
economists to classify children as "inferior goods," at
least over a substantial range of income. However, this
does not mean that child care exhibits the same
relationship to income. Economists from the Chicago
school34 have argued that the number of children per
household (N) and the amount of expenditures per child, or
quality (Q), serve as substitutes in the household welfare
function W = w(N, Q, S), where S is a composite measure of
all other household commodities, indicative of the
household's standard of living. The reason that N and Q
are hypothesized to be substitutes rather than complements
is because the price of an additional child (Pn) is greater
the higher the expenditures per child (Q) of existing
children. Conversely, the price of a per unit increase in
quality (Pq) is higher the greater the number of
children.35 Therefore, Pn = pQ and Pq = pN, where p is the
marginal opportunity cost of the composite output of child
services. The relative marginal costs of Pn and Pq is
pQ/pN, or simply Q/N. Unless Q and N change in equal
proportions, the relative marginal costs of Q and N will
47


change given a change in expenditures on child services.
Willis36 assumes complementarity between the parent's
standard of living (S) and expenditures per child (Q) ,
creating an initial bias toward purchases of quality
resulting from an increase in income. Since the price of
an additional child is positively related to the level of
quality per child, the higher level of quality will
thereafter increase the price of any additional child.
Although the resulting income elasticity for number of
children may not be negative, it is presumed to be very
close to zero. This is part of the explanation for why
average real incomes in the U. S. have risen more than
threefold in the last few decades while fertility rates
have tended to decline, making it appear as if children
were "inferior goods." Under the household production
model, we would predict that quality of child care, or the
amount of care given a child, would rise with income,
rendering the composite output of child services a normal
good. And since center care is a mode of child care; we
would expect a positive coefficient for the income
variable.
However, from the same general model, there is an
alternative hypothesis generated in the case where the
increased household income represents the husband's
earnings. In this case, the household production model
48


predicts a greater tendency for the wife to devote more
time to household workincluding child rearingsince the
value of the marginal product of the female's nonmarket
time is now higher. In other words, with a greater supply
of the input of "market goods" into the household's
production function, "own-time"or time spent working
outside the labor marketis now relatively more scarce and
therefore entails a higher marginal product, dictating its
increased employment as an input. Intuitively, the
argument is simply depicting the efficiency of the
"traditional" American family with its division of labor
between husband and wife when the husband holds a
comparative advantage in the labor market. In this case,
a higher income would find a tendency for the wife to
devote more time to the home and provide child care
herself, rather than purchasing the care from a center.
Thus, we might expect to find a negative coefficient for
income if this scenario was plausible.
This scenario, however, is not in accord with the
historic trend of women entering the labor market in spite
of rising real per capita incomes over the same period.
Cultural changes may be having an impact, but at least part
of the reason is that higher incomes have meant time-saving
household products which serve to increase the supply of
the female' s nonmarket time and reduce its value at the
49


margin. At the same time, female wages have been rising as
employers respond to the rising wage demands of the
traditional male labor force, and the wage elasticity of
demand for labor is generally higher for lower wage rates,
meaning that women have responded to the offers of higher
wages even though they are still not on par with the
earnings of men. Moreover, as a woman spends more of her
time in the labor market, her average wage may be expected
to rise. With these considerations taken into account, we
may interpret the increase in real income over the past few
decades as in part "causing" the emergence of women into
the labor force as well as being "caused" by their
increasing participation. This interpretation is
significant because part of that 325% increase in real per
capita income is now attributed to higher female wage
rates, which greatly increase the opportunity cost of
bearing children since it is the female's time that is a
critical factor in the cost of bearing and rearing young
children. Now there is a much greater substitution effect
entailed by higher income, serving to limit the number of
children while inducing greater purchases of care outside
the home by a working female. Once again we would expect
a positive coefficient for the income variable noting the
female's incentives to enter the labor market rather than
to remain at home caring for her children.
50


Title
X2: Fed/state Title XX/SSBG per capita.37
XX/SSBG is a major source of Federal and State funding for
child care, apart from the child care federal income tax
credit, which should be reflected in other variables,
namely income and birth rates. Title XX is important for
picking up the subsidy targeted to lower income households
which do not benefit from a nonreimbursable tax credit.
Per capita subsidy should be correlated positively with the
dependent variable.
X3: The difference between daily center reimbursement
rates and FDCH daily reimbursement rates.38 This difference
is set by the designated state authority. Calculated by
subtracting FDCH rates from center rates, the resulting
number was either zero or positive, indicating that
reimbursement rates were always higher for centers when
there was a difference. A larger difference should be
positively related to a higher equilibrium quantity of
center care, represented either as a shift outward of the
-supply curve of center slots, or a shift outward of the
demand curve for center slots, depending upon how the
subsidy is administered in the state.
X4: Female labor force participation rates.39 It is
expected that states with higher female labor participation
rates should exhibit a higher rate of consumption of center
slots, assuming of course that women with small children
51


make up a fraction of that higher percentage. In other
words, the higher statistical rate of participation is
being read as a greater propensity for the average woman to
work, all else being equal.
X5: The percent of population living in urban areas.40
Centers may be expected to have advantages in locating
within areas of relatively dense residential populations,
thereby assuring a large, diverse market with a minimum of
transportation costs. Minimizing transportation costs
confer external benefits upon all child care establishments
located in urban areas by shifting their long run average
cost curve downward. It is possible that these external
benefits also change the shape of the long run average cost
curve, allowing for a greater range of decreasing costs
over output, thereby enabling establishments to expand and
exploit these economies of scale that were essentially
negated in less dense areas by the increasing
transportation cost for each additional child drawn from a
larger area. A positive coefficient is anticipated for
this variable.
X6: The percent of births to mothers with less than
high school education in 1984.41 This variable is intended
as an approximate measure for the average educational
attainment of females across states, indicated by the
percent of females with less than high school education.
52


Economists42 have hypothesized several links between the
female's educational attainment and the level of child
services in the household. For example, the educational
attainment of the female may directly affect the family's
preferences for number of children, mode of child care, or
both. Education may affect knowledge and efficiency in
birth control and family planning. Education may also
affect the productivity of the female's time in both the
labor market and household production activities, thereby
changing the female's market wage rate as well as her own
price of time when she is not in the labor force. The
most often cited link would be changes in the female's
market productivity and her expected market wage, wherein
more education translates into higher wages. In this case,
a lower female wage statewide would be expected for higher
values of X6. This would impact the equilibrium quantity
of center care from both the supply and demand side. From
the supply side we would expect to see a shift outward in
the supply curve due to the greater supply of labor for day
care centers. This would lower price and increase the
quantity demanded of center care.
On the demand side the linkage is a bit more tenuous,
but otherwise valid. Since rearing of young children is
thought to be female time intensive relative to other
household activities, a higher female market wage would
53


mean a higher opportunity cost of producing child services.
Low levels of education would therefore mean lower foregone
market wages and a lower opportunity cost of bearing
children. Therefore, females with relatively low
educational attainment should exhibit a higher propensity
to bear children. And while households may spend more on
child services without having more children, more children
necessarily means greater expenditures on child services,
this in spite of the fact that the price elasticity of
quality may be negative, zero, or even slightly positive
while still allowing the price elasticity for the composite
output of child services to be negative, since Q represents
expenditures per child, not total expenditures.43
Therefore, even from the demand side we would expect to see
a greater equilibrium quantity as a result of the demand
schedule for center care shifting outward. The ambiguity
is in the equilibrium price, not quantity. In either case
we would expect to find a positive coefficient for this
variable.
X7: Per capita birth rate for 19 8 3.44 The measure of
female educational attainment (X6) may be confounded by the
tendency for less educated women to have more children,
thereby over-representing the number of women with less
than high school education out of the population of all
women in the state. If this over-representation is
54


consistent in direction and magnitude across all states,
then the variable would still be a good indicator of
average educational attainment. The problem, however, is
the presence of other factors, such as the husband's
earnings, that systematically affects the less educated
female's decision to bear children.
For example, if the presence of a husband is more
likely in some states or if his average earnings are
higher, then females with less than high school education
may be less likely to bear children than those women with
equivalent educational attainment in other states, since
the husband's income constitutes a greater opportunity cost
of bearing children in terms of the household's standard of
living, rather than the female's market opportunities. In
these states, the percent of females with less than high
school education would be underestimated, due to the lower
likelihood of their being within the subcategory of
female's giving birth in 1984. In the absence of these
mitigating factors, we would, however, expect to see a
positive correlation between this variable and X6, in
accordance with previous data indicating a negative
relationship between the female's educational attainment
and birth rate. Inclusion of this variable is therefore a
weak test of this relationship. Moreover, if there is such
a relationship, we would expect a positive coefficient for
55


this variable.
Xg: A dummy variable for states with prekindergarten
legislation.45 For this variable 1 = legislative programs
intact, and 0 = no such legislative programs.
Prekindergarten programs may be considered as substitutes
for centers, although not very good substitutes, since they
are usually targeted for specific segments of the relevant
population and have limited hours of operation in the day.
The more comprehensive and generously funded state
prekindergarten programs, however, should serve to decrease
demand for center care, all else being equal. In 1987,
eleven out of the twenty-three states included here as
having preschool legislative programs had expanded
preschool program hours and days in an attempt to meet the
demands of the child care market.
X9: The state's licensed child/staff ratio for three
year-olds.46 This is the first of the two policy variables
included in the regression. That quality child care may be
significantly correlated with lower child/staff ratios is
not difficult to fathom. The more children a staff worker
must attend to, the less attention any one child may
receive from the staff member. Moreover, the ability of
the staff member to provide quality care may be adversely
affected by the fatigue of attending to more children. The
correlation between child/staff ratio and quality of care
56


need not be strictly linear, but there is little doubt that
after a point quality care must significantly decline with
additional children assigned to the staff member.
Moreover, that for-profit centers may settle on a
child/staff ratio in excess of this optimum ratio is also
likely, since overhead as well as operating costs can be
reduced by placing one more child under the care of a staff
member. Hence, the child/staff ratio presents itself as
one of the logical conditions to focus on in attempting to
raise the quality of child care.
As an important indicator of quality in its own right,
this policy variable will impact both supply and demand
curves. Raising the licensed child/staff ratio (i.e.,
allowing more children per staff member) would be
indicative of lower quality, and we would expect the demand
curve to shift downward. However, this will also impact
the cost curve for centers. Raising the child/staff ratio
would allow centers to increase the number of children
served by one staff member while holding constant the
revenue per child, thereby reducing average costs and
increasing profits as well as the supply of licensed center
slots at all prices. Furthermore, it might be expected
that an increase in the licensed ratio from, say, 8
children per staff member to 9 children would have more
impact in terms of lowering average costs than an increase
57


from, say 14 to 15 children per provider, since fixed costs
decline at a decreasing rate as output (children served)
increases/7 For this reason, the square root of the ratio
will be tested as an alternative functional form.
The net effect will reflect the relative magnitude of
the shifts in' the supply and demand curves, and therefore
will reflect the relative magnitude of the costs and
benefits associated with a change in the licensed
child/staff ratio. A positive coefficient for this
variable would indicate net benefits to society for raising
the licensed child/staff ratio (i.e., allowing more
children per staff member), while a negative coefficient
would indicate net costs associated with raising the
licensed child/staff ratio.
X1Q: An ordinal measure of preservice requirements for
directors of centers/8 Preservice educational
requirements is the second policy variable chosen for the
regression. The most likely rationale for requiring
educated providers is due to the supposition that much of
what may be considered higher quality in a day care program
has to do with the relationship between staff and children,
and if love cannot be quantified and mandated for staff
members, then knowledge may do as a second best. There is
some evidence that staff members more knowledgeable in
child development enhance child/staff relationships/9
58


Preservice requirements in childhood development or
early childhood education are greatest for directors or
head teachers in day care centers, while teachers and aides
rarely require more than a high school education and a few
hours of orientation.50 This may seem inefficientsince
it is the teachers and aides who are most frequently in
contact with the children-but in fact may not be. The
presence of at least one full time specialist may suffice
to improve the quality through continued supervision,
training, and selection of staff members by the specialist.
In effect, on the job training" may be a more effective
method of early child education and development as well as
a less costly method.
More importantly, the director is the one responsible
for maintaining the licensed standards of the center, and
focusing this responsibility on one person will avoid the
"free-rider" problem associated with the production of what
is essentially a public good, i.e., an environment
reflecting some minimum quality of care. Therefore, even
though the license in this case pertains not merely to the
provider but to the day care "facilities,"which Class and
Binder define as the "...people, operations, structure and
materials,"51it is the director who stands to lose the
most from failing to uphold the licensing standards. In
this respect, the director of the day care center bears a
59


responsibility similar to that of the physician in
Svorney's article, i.e., as the "residual claimant" in the
provision of services.
The measure of preservice requirements adopted in this
paper comes directly from Gwen Morgan's52 1986 study.
Using essentially the same categorization of states as
Morgan does for representing a state's preservice
educational requirements for directors, 1 = no preservice
requirements, 2 = experience only or substitutable for
education, 3 = some college and usually some experience
along with it, and 4 = substantial college, with early
childhood education or child development as part of the
degree.
Clearly, a higher value will shift the supply curve
inward. However, if the added value of higher requirements
are relatively large, then the outward shift in the demand
curve may exceed the inward shift of the supply curve. If
this is so, then these preservice' requirements benefit
consumers as well as providers, and a positive coefficient
would be expected.
The problems associated with using an ordinal variable
are recognized, and for this reason different functional
forms were allowed, including the square root of the
variable as well as the variable squared.
60


Summary of Chapter
The observed disparity in licensing requirements
across states for child care centers may be explained in
several ways. One possible explanation is that there is
considerable confusion over what licensing is supposed to
accomplish and how it may accomplish its desired effects in
any given industry. The models developed in this chapter
are intended to provide a relatively simple method of
ascertaining the net effects of licensure in an industry
such as child care. The theoretical model illustrates the
effects of changes in the key variable (W) in equation
(2.5) on the supply and demand schedules for center care as
a result of raising licensing standards in this market.
The empirical model then translates these effects into
measureable variables included in a regression equation.
61


CHAPTER 4
RESULTS AND DISCUSSION
The results of the specified regression can be seen in
Table 4.1 below. An unadjusted R2 of 0.68 was obtained,
with an adjusted R2 of 0.57. All but four of the variables
were significant at the ten percent level for a two-tailed
test.
Table 4.1 . Regression Results
VARIABLE COEFFICIENT STD. ERROR T-STAT. 2-TAIL
SIG.
C -0.0437202 0.0157187 -2.7814074 0.010
X2 0.0003953 0.0002939 1.3450710 0.190
X3 0.0005042 0.0004351 1.1587755 0.257
X4 0.0005553 0.0002124 2.6151405 0.015
X5 6.388E-05 3.599E-05 1.7747154 0.088
X6 0.0005986 0.0002055 2.9134865 0.007
X7 -0.1488671 0.3644734 -0.4084442 0.686
X8 -0.0019500 0.0017894 -1.0897308 0.286
X9 0.0049845 0.0028609 1.7422970 0.093
X10 -0.0024032 0.0008822 -2.7241062 0.011
Notes: R2 = 0.68 Adjusted R2 = 0.57
Some observations excluded because of missing data.
Number of observations =36
A covariance matrix for all independent variables was
also run, with the results shown in Table B.l in Appendix
B.


Non-Policv Variables
Before the policy variables X9 and X10 are discussed,
a brief discussion of the the non-policy variables will
follow, beginning with income and proceeding in the order
of appearance in the equation.
The income variable was dropped since it showed signs
of collinearity with several other variables, including per
capita subsidy, female labor participation rates, percent
urbanization, and percent of births to mothers with less
than high school education. In fact, a separate regression
with per capita income as the dependent variable and X2, X4,
X5, X6, and X7 as independent variables was run resulting in
an adjusted R2 of 0.72 and all independent variables
significant at the 10% level (see Appendix C) Omission of
the income variable in the primary regression did not
significantly change the coefficients of the remaining
variables, while it did improve most of the t-values as
well as the overall fit of the equation to the data.
The coefficient for per capita subsidy was not
significant at the 10% level. This may be due to the fact
that many for-profit centers do not participate in child
care subsidy programs. Moreover, there may be a crowding
out effect in those states with higher per capita subsidies
if these centers find it difficult to compete with non-
63


profit subsidized centers willing to forego what is
considered a normal, rate of return on their investments.
A regression run with the income variable included resulted
in a much lower level of significance for the subsidy
variable, indicating that it is picking up the effects of
a state's per capita income and potential for tax revenue.
The coefficient for the difference in rates for
centers and FDCHs likewise was not significant at the 10%
level but was kept in the equation even though omitting it
resulted in a higher adjusted R2 (see Appendix D). Theory
indicates it probably belongs in the equation and its
omission changes the coefficients of the remaining
variables.
The coefficient for female labor force participation
rate was positive, as expected, and significant at the 10%
level.
The coefficient for percent urbanization was positive,
as anticipated, and significant at the 10% level. This
variable also appears to be picking up the effects of
higher income on the demand for center care, perhaps even
to the extent that the hypothesized impact on the supply of
center care through beneficial externalities and economies
of scale is not significant.
The coefficient for the percent of births to females
with less than high school education was positive and
64


significant at the 5% level.
The coefficient for birth rates was negative but not
significantly different from zero. The negative
correlation between birth rate and income plus the positive
impact of income related variables on the demand for center
care lend support to the quantity/quality tradeoff
hypothesis. However, the statistical nonsignificance of
the simple correlation between birth rate and percent of
births to females with less than high school is not in
accord with predictions that lower female education
translates into higher birth rates. Even with mitigating
factors left out (such as husband's average earnings) we
should have seen a positive correlation, if the female's
education is to have any impact at all. What might explain
this result in the face of prior evidence supporting such
a relationship? One possibility is that the level of
aggregation is simply too great to pick up the relation.
An analysis with counties or even individual households as
the units of analysis would be more appropriate. Whatever
the case, the positive influence of this variable is
attributed to the supply side. In other words, it is the
supply curve rather than the demand curve that is shifting
outward due to the greater supply of potential child care
workers, i.e., mothers with less than high school
education.
65


The coefficient for preschool legislation likewise
came in negative, as anticipated, but was not significant
at the 10% level.
Policy Variables
Now the focus of the discussion will turn to the
impact of the two policy variables on equilibrium quantity
exchanged, beginning with the child/staff ratio.
The coefficient for child/staff ratio came in
positive, and was significant at the 10% level. The square
root of the licensed ratio was tested as an alternative
functional form and favored over the unadjusted value on
the grounds that it improved slightly the overall fit of
the equation and was theoretically justified. Taking the
square root also had the effect of more than doubling the
variable's coefficient (along with the standard error),
thereby increasing the impact of licensed ratios on
equilibrium quantity.
The positive sign for the first policy variable,
child/staff ratio, indicates that the lowering of costs
exceeds the decrease in value for the consumer when the
ratio is raised over the range of child/staff ratios
existing for three year-olds.
Finally, the coefficient for preservice requirements
came in negative, and was significant at the 5% level.
66


Different functional forms, including the square root of
the variable and the variable squared were tested but did
not improve the overall fit of the equation. In the
absence of any other plausible hypothesis the ordinal value
was employed.
The negative sign for the last policy variable, X10,
indicates that the added value of the requirements is less
than the added costs of educating directors. While
preservice licensing requirements for directors may be in
the interests of some, it is generally not in the interests
of consumers.
These results do not bode well for those who believe
that stricter licensing requirements will benefit consumers
of child care by raising standards for quality. The
question now is why?
Beginning with the assumption stated in the
Introduction that qualityas judged by the consumeris
not lowered by raising licensing standards and is
presumably raisedalbeit perhaps even slightlythe
reasons for the observed signs of the policy variables can
be divided into two broad categories, the first focusing on
arguments regarding the relative efficiency of licensing as
a means of raising quality or lowering information costs,,
and the second focusing on the individual household's
"willingness" to purchase quality.
67


Child Care as a Public Good
Taking the second category first, the meaning of
"willingness" to purchase must first be clarified. Taken
at face value, the signs of the coefficients indicate that
parents do not see the higher quality as being "wcprth it."
However, some may claim parents in fact value the service
at the higher price but reduce purchases because they
simply cannot afford it. In other words, they are
"willing" to purchase higher quality care but are unable
to.
But what does this really mean? We may value a new
car at its full price (we say its "worth it") but decline
to purchase it. The rationale we give is that we simply
cannot afford it. What we mean by this is that if we had
more than we have now of everything else (i.e., higher
income) then we might be willing to give up the same
quantity of goods and services to acquire the car since the
quantity we give up now has a lower marginal utility. The
rationale is exactly the same When we decide not to buy the
car even though we could "afford" tothe marginal utility
we lose from the goods and services we give up is greater
than the marginal utility we gain from the car we purchase.
In either case, when we say that we value the good at its
market price we mean that it is not being produced above
cost, or that its price reflects efficient means of
68


production and that we would not expect to find the same
quality for a lower price. If we do not purchase it this
means we are not willing to purchase it at that price
nothing more, nothing less.
Unless the stricter requirements for center care do
not result in higher quality service it makes little sense
to say that parents are willing to purchase the same amount
but cannot afford to, as the same may be said of all goods
and services currently not sold on the market. What those
who lament the affordability of quality child care mean to
imply is that all parents, regardless of their income,
"should" receive some minimum amount of quality care deemed
necessary by society.
Certain goods and services can be thought of as
constituting those society deems necessary in one form or
anotherfood, clothing, and shelter, for example. These
are goods that ordinarily exhibit income inelasticity and
consequently consume an ever increasing percent of one's
expenditures as income declines. If these goods also
exhibit the composite attributes of quantity and quality,
then it may be expected that the income elasticity of
quality exceeds that of quantity, just as in the case of
child services. In other words, as income declines, the
quality of these goods declines more than their quantity.
Intuitively this means we may live in a one bedroom motel
69


room rather than a one bedroom house, commute about the
same number of miles each day but in a bus rather than car,
or own about as many clothes but all purchased from a
second hand store. At very low levels of income, the
quality of these goods and services may decline to such an
extent that they are well below society's cultural
standards, thereby creating a "second class" citizenry.
A minimum level of quality may be mandated by society
for these goods on the rationale that all citizens merit
them as members of society. Ensuring consumption of a
minimum level of quality goods and services by lower income
classes is justified if there are significant external
benefits associated with this consumption. The potential
benefits that accrue to those who must subsidize the
consumption of these basic services by others is in
avoiding a widely disparate standard of living which could
ultimately disrupt society. If these external benefits
exist, then the good or service under consideration
exhibits characteristics of a public good. If child care
is one of these "basic" services, then child care may be
considered a public good.
Besides positive externalities of consumption, child
care may also exhibit positive externalities in production.
It is possible that in the past parents relied heavily upon
their offspring for the services they provided later in
70


life, but that now these services are provided through
private pensions and social security benefits. Though it
is still the younger generation who indirectly provides
these benefits through their productivity, now the
offspring provide, in effect, for many households entering
retirement, and not just their own. If the benefits of
early childhood education are increasingly appropriated by
society in general and not the family in particular, then
the parents will invest too little in their preschool
education when it is left as a private decision. Parents
are unwilling to purchase higher quality care in this case,
and even though their unwillingness is rational from their
standpoint, it is not rational from society's standpoint.
Of course, the weight of these arguments that regard
child care as a public good rests upon the degree to which
experiences in the first six years of life significantly
affect the child's overall social adjustment later in life
and whether public intervention will actually improve this
adjustment. There is some evidence from programs like Head
Start53 that quality care outside the home for preschoolers
from low income families will have lasting beneficial
effects. However, the impact on children from middle class
families is less clear cut.
71


Licensure as a Means of Ensuring Quality Provision
Even if child care is a public good requiring
subsidization, this is not necessarily an argument in favor
of stricter licensing requirements. The reason is that
licensure relies upon the individual's marginal valuation
of the good or service rather than society's marginal
valuation.
But now suppose that child care is not a public good.
The regression results still do not necessarily imply that
the average individual household is unwilling to pay for
higher quality child care.
Recall that part of the benefit consumers receive from
licensure is the added assurance of higher quality, i.e.,
a lowering in the cost of ascertaining the quality of any
particular firm, which includes the costs associated with
the consumer's own research and. with monitoring the
service. We have adopted the assumption that the
licensing requirements do in fact raise quality if they are
adhered to. However, because of the added assumptions that
quality provision is a decision variable and that there are
increasing marginal costs for quality provision in the
child care market, the actual quality provided through the
operating life of the firm may still not be the potential
quality, because the license may not serve to eliminate or
significantly lessen the incentive to reduce operating
72


costs by reducing quality. This implies that the "wealth"
effect is not operating, due to non-optimal values of W, S,
or p in the incentive function
I=E[ (1-p) U{W+w) +pU{W+w-S) ] -U(W) (4.1)
previously defined in Chapter 2. Recall that so long as
J(-)>0 (4.2)
there is a positive incentive to lower quality. Because of
the assumptions of information asymmetry, increasing
marginal costs of quality provision, and quality provision
as a decision variable, the incentive function is presumed
to be positive in the child care market. However, since
(4.3)
and
(4.4)
under most conditions, these variables can be manipulated
by licensing authorities attempting to improve quality in
the industry, subject of course, to their budget
constraints. But another set of constraints involves the
actions households and.firms may take in response to these
73


interventions in the market. The following examines these
two constraints.
Assume first of all that standards are enforced and
the probability of being detected (p) for providing
substandard quality is high. However, the deterrent effect
of the sanctions (or the value of S) may simply not be
effective. If there is no jail term or criminal fine, then
the harshest sanction will be revoking a director's license
and possibly the licenses of teachers in the program. But
the loss of wealth in this case is limited by the
opportunity cost of the particular provider. The more
specialized is the provider's training, the greater the
loss.
In the case of licensed center care, what undermines
the efficacy of sanctions is the presence of the largely
unregulated market for day care, namely, family day care
homes (FDCHs). Recent surveys54 indicate that the
differential in the price- of center and FDCH service is not
very large, hence, neither would the difference in net
returns be large if a network of FDCHs or a group home is
operated. Licensing only a fraction of a.market or only
one of two closely related markets would not seem
particularly effective unless those whose licenses are
revoked are explicitly forbidden to operate a FDCH.55
On the other hand, if lack of enforcement (a low value
74


of p) has become general knowledge then parents will not be
willing to pay more for staff requirements when other
requirements are not being enforced, or when the sanctions
applied do not have the intended deterrent effect.
Removing enforcement and the likelihood of being punished
for poor quality service essentially lowers the provider's
expected costs associated with a reduction in quality,
increasing the likelihood that he will reduce quality,
especially if the director is under pressure to repay or
recoup the educational expenses incurred in meeting the
requirements.
For example, consider the potential ineffectiveness of
raising educational requirements for directors when there
is already a tendency for the marginal center (the one with
highest operating costs) to skirt existing requirements.
Suppose also that demand is fairly sensitive to price (not
an unrealistic assumption for center care). Faced with the
loss of revenue if prices were raised, directors of for-
profit centers entering the market could cut costs in other
areas. But the only other costs worth considering would be
staff salarieswhich constitutes the majority of costs
and overhead costs.
Most of the overhead costs borne by a center are
mandated by other licensing requirements, and if the
expected loss of skirting these requirements is lower than
75


the expected loss of raising prices to fully cover the
director's educational expenses, then overall quality may
not increase, and may even decline. Although this does not
violate the assumption that raising licensing standards
will not lower quality (since the standards are no longer
being adhered to) the cpnsumer does not see it this way.
Alternatively, the director could lower staff
salaries. But child care workers are already receiving low
pay compared to similar jobs in other industries, and a
decline in staff wages would further exacerbate the problem
of high turnover, low morale, and low quality.
Other directors may choose to cut costs by allowing a
higher licensed child/staff ratio. In fact, more educated
staff in return for higher child/staff ratios is the
tradeoff made in the French child care system. The
positive sign of the coefficient for child/staff ratios
indicates this could be a viable tradeoff. However, the
sign could quickly change once higher child/staff ratios
are encountered, as the decline in quality may eventually
exceed the savings in costs passed on to the consumer. In
fact, it is possible that centers refrain from increasing
their child/staff ratios due primarily to the presence of
family day care homes as substitutes, rather than from any
licensed requirements. If this is the case, then centers
may well experience a decrease in demand associated with an
76


increase in child/staff ratios after a particular point,
resulting from a relatively large shift inward of the
demand curve for center care. This implies that diversity
in the child care market may be an important characteristic
in maintaining quality, as long as parents are capable of
exercising a variety of options in the day care market.
Although all these responses directors could make in
the face of higher standards might imply that tougher
enforcement of all licensing requirements could lead to
greater assurance for consumers and a net benefit to
society, this argument does not take into account the
higher expected costs associated with tougher enforcement.
For example, if the degree of enforcement is directly
related to the costs of enforcement, which in turn depends
upon the number of establishments covered by the license,
the number of separate government agencies responsible for
enforcing requirements, the complexity of the licensing
requirements, and the number and qualifications needed for
the licensing staff, then the higher costs of enforcement
may soon outweigh any added benefit of quality assurance
for the consumer. Now the budget constraint of the state
comes into effect.
Finally, there is the very real possibility that the
"legitimate" wealth (W) or the salary of a licensed
provider, is simply too low to discourage them from
77


lowering quality and reducing costs. Raising salaries by
setting more stringent standards may ensure quality, but it
will raise net benefits to society only if most of the
households value higher quality and an additional hour of
their own-time is expensive relative to an additional hour
earning income. If this is not the case, then the license
will only serve the interests of those few providers left
in the market for licensed center care. In this event,
subsidizing staff salaries rather than raising licensing
\
standards would prove more effective, but only if the
(increased) subsidization could be justified by arguments
similar to those in the preceding section.
Summary of Chapter
To summarize, results of the regression analysis
indicated that households are apparently unwilling to
support higher licensing standards. This could mean either
that households are generally unwilling to pay for higher
quality, or that households do not believe licensing
actually raises quality substantially or at a price that is
lower than what it would cost a typical individual securing
higher quality in the market on their own.
If individual households are unwilling to pay for
higher quality (as defined by the licensing authorities),
then perhaps quality child care exhibits features of a
78


public good, thereby requiring greater public financing.
If this is the case, then raising licensing standards will
not achieve the desired results because it inherently
relies upon individual valuations of what is essentially a
public good.
If, on the other hand, parents in general are willing
to pay for higher quality, then a more cost effective
method of ensuring quality apparently must be found.
79


CHAPTER 5
CONCLUSION
The summary of the preceding chapter essentially
contains the conclusions of the research as part of the
discussion of the policy variables and the observed signs
of their coefficients. Those comments will not be repeated
here; rather, this conclusion will draw upon or extend the
discussion of the results to considerations of alternative
means of raising guality in the child care market for the
least cost.
Consideration of the two broad explanations for the
signs of policy variable coefficients in the preceding
chapter serves as a guide to the discussion of the possible
alternatives to using licensing requirements to raise
quality standards.
First of all, if the requirements are not being
adequately enforced, then this is an indication that the
costs of enforcement are perceived by agencies as being
prohibitive. The alternative is not eliminating licensing
per se, but a simplification of licensing requirements and
the method of enforcement. The first step is perhaps to
recognize the limitations of licensing, as Gwen Morgan56


points out. The license should serve to maintain a minimum
standard for all or nearly all modes of child care outside
the home, rather than a mere fraction of this care. The
minimum standard could be defined in terms of the estimated
value of the care provided a child by a mother in the home
full time with an income somewhat above the official
poverty line. How far above the poverty level the standard
is set depends upon the effort officials make in reaching
those children who are cared for in impoverished homes, for
it makes no sense to legislate quality care in the market
while allowing many children to receive lower quality care
at home.
The simpler and more streamlined the requirements the
more establishments may be covered. It may still be too
costly to include many of the smaller family day care homes
scattered throughout the city, but the option of purchasing
minimum standard inexpensive care at a licensed center
might serve as a viable substitute to FDCHs and would
therefore help to keep FDCH standards in line with those of
centers.
For higher standard care, or even for defining what
higher standards might be, voluntary accreditation of child
care providers and/or certification of child care programs
and facilities, combined with the minimum licensing
requirements, is probably preferable to stricter licensing
81


requirements. The idea is to encourage creativity and
experimentation in early childhood programs rather than in
attempting to enforce a rigid standard presumed to
represent the best and only way to care for preschoolers.
At the same time, maintaining minimum standards through
licensing may serve to define the boundaries of what is
socially acceptable in the range of child care programs
that is likely to surface. Through voluntary accreditation
or certification standards could be raised in many
establishments without forcing minimum standard centers out
of the market.
For credentials to serve as quality assurance signals
they should be established and recognizable. The Child
Development Associate credential is one possible example.
Endorsements may be awarded by high profile organizations,
such as the National Association for the Education of Young
Children (NAEYC), the American Federation of Teachers
(AFT), the Children's Defense Fund (CDF), the Child Welfare
League, and even the American Psychological Association,
although many such organizations may only lend their name
to large service chains rather than a single establishment.
A variety of such credentials and endorsements, tailored to
meet the specific goals of the program, would reflect a
range of child care programs serving different segments of
the market with different needs.57
82


For credentials to have meaning, and as a check for
abuse of the system, a public or semi-public information
and referral system may be established with the aim of
matching the needs of parents to the available programs in
the area. Among the tasks of an information and referral
system would be providing information to parents on where
to find care, what to look for in a quality center, a
profile of existing establishments, and a place to register
complaints or commendations. The referral agency should
not be "captured" by any one child care establishment or
group of establishments, and so therefore should not accept
fees or advertisements from the industry, nor should they
be directly involved in inspecting the centers.
Information and referral centers could also help to educate
young parents and train new staff members by providing
discounts on hospitals or universities offering such
courses or seminars.
Minimum standard licensing, voluntary accreditation,
and information and referral are institutional arrangements
designed to maximize consumer choice in markets with
asymmetry of information or uncertainty. But to the extent
that consumers are knowledgeable and yet their marginal
private benefit (cost) is not equivalent to the marginal
social benefit (cost) of child care provisions, then
greater public provision or subsidization may be called for
83


if the costs of increased public intervention do not exceed
the differential in social and private benefits.
If the differential between private and social
benefits is greatest for lower income families, then the
bulk of public assistance should be targeted for these "at
risk" groups. This is apparently the intent of the Budget
Reconilication Act of 1990, which not only increases block
grants to states for funding of child care services, but
also expands the earned income tax credit for low income
working families with children.
The Reconciliation Act also expands public provision
of child care in effect by increased funding for Head Start
to accommodate more eligible children. Indeed, it is
conceivable that Head Start or a similar national preschool
program (possibly utilizing the public schools) would even
replace the need for minimum standard licensing in most
states. Of course, this would only be the case if the
scope of operation of this preschool program is such that
it suits the needs of working parents. Even if this is the
case, however, studies comparing the costs and benefits of
such an expansion with the costs and benefits of
establishing and enforcing, minimum standards would be
called for.
84


APPENDIX A
LELAND'S FORMAL MODEL OF THE EFFECTS OF INFORMATION
ASYMMETRY ON THE PROVISION OF QUALITY
Let q equal the percentile of potential supply which
has quality q or less, so that 0 < q < 1. Let q equal the
highest quality supplied and q equal the average quality
supplied, or
(A. 1)
By assumption, the marginal cost of supplying quality
is positive, i.e.,
d^g) >0 (A. 2)
dq
Then the supply price Ps will be equivalent to C'(q)
since all suppliers with q < q will sell at Ps and all
suppliers with q > q will not. Therefore
q=y (A.3)
Let marginal willingness to pay (or the inverse demand
curve) equal Pd where
Pd=P(q,y) (A. 4)
where
P
(A.5)
By substitution,


The equilibrium price is
C(3e)=p(-|3e,£e)=Pe (A. 7)
equilibrium supply is
ye= and equilibrium average quality is
u-9)
Let W = net benefits = consumer surplus + producer
surplus, or
W=fp( Substituting for y and q from (A.3) and differentiating
w.r.t. q yields
-^ = f%(<3,y')dy'+p(q,$)-C(3) (A. 11)
dq 2 J
At q = qe = ye implies
-^U=^{%g(^y')dy'>0 [A. 12)
That is, since (A.12) is positive, at equilibrium net
benefits will increase with increases in quality.
86


APPENDIX B
THE COVARIANCE/CORRELATION MATRIX
Table B.l. Covariance/Correlation Matrix. Series Mean S.D. Maximum Minimum
XI 10388.732 1790.0157 14882.000 7421.0000
X2 3.7254582 2.9935868 14.647620 0.0000000
X3 1.6814634 2.0130456 9.5800000 0.0000000
X4 57.129268 4.9690665 65.800000 40.400000
X5 62.651219 21.689330 100.00000 23.100000
X6 20.048780 6.4805525 37.000000 9.0000000
X7 0.0159341 0.0027046 0.0252000 0.0128000
X8 0.5121951 0.5060608 1.0000000 0.0000000
X9 3.3930785 0.3524143 4.0000000 2.6457510
X10 2.4390244 0.9232762 4.0000000 1.0000000
Series Covariance Correlation
XI, XI 3126006.0 1.0000000
XI, X2 2587.8260 0.4950057
XI, X3 1073.9011 0.3054764
XI, X4 4326.8056 0.4986084
XI, X5 22323.518 0.5893639
XI, X6 -6032.3040 -0.5330135
XI, X7 -1.3232616 -0.2801634
X1,X8 304.57644 0.3446360
XI, X9 -142.17410 -0.2310117
X1,X10 316.50803 0.1962999
X2 X2 8.7429873 1.0000000
X2,X3 0.2369885 0.0403093
X2,X4 2.3067588 0.1589496
X2 X5 14.303371 0.2258002
X2,X6 -5.3088911 -0.2804941
X2,X7 0.0011036 0.1397153
X2,X8 0.6654350 0.4502305
X2,X9 -0.1679712 -0.1631974
X2,X10 -0.2740280 -0.1016238
X3,X3 3.9535149 1.0000000
X3,X4 1.0424445 0.1068190
X3,X5 7.5448266 0.1771224
X3,X6 -4.2237299 -0.3318596
X3 ,X7 -0.0014539 -0.2737141
X3,X8 0.1509578 0.1518878
X3,X9 0.1668992 0.2411409
X3,X10 0.4439917 0.2448575
87


Table B.l. (cont'd)
Series
Covariance
Correlation
X4,X4 24.089387 1.0000000
X4,X5 9.1455738 0.0869789
X4, X6 -20.838013 -0.6632745
X4,X7 0.0024475 0.1866711
X4,X8 -0.4052350 -0.1651784
X4,X9 -0.1768997 -0.1035434
X4,X10 0.4017845 0.0897657
X5,X5 458.95321 1.0000000
X5,X6 -9.7854238 -0.0713584
X5,X7 -0.0149569 -0.2613468
X5,X8 3.3420584 0.3120966
X5,X9 1.1651799 0.1562490
X5,X10 5.9823910 0.3062111
X6,X6 40.973230 1.0000000
X6,X7 -0.0030261 -0.1769650
X6,X8 -0.2688876 -0.0840388
X6,X9 0.6836071 0.3068068
X6,X10 -1.2897085 -0.2209387
X7,X7 7.136E-06 1.0000000
X7,X8 -0.0004711 -0.3528395
X7,X9 0.0001947 0.2093852
X7,X10 -0.0004613 -0.1893673
X8,X8 0.2498513 1.0000000
X8,X9 -0.0152874 -0.0878618
X8,X10 -0.0053540 -0.0117453
X9,X9 0.1211666 1.0000000
X9,X10 0.0385129 0.1213235
X10,X10 0.8316478 1.0000000
Note: Some observations excluded because of missing data.
Number of observations =41
88


APPENDIX C
REGRESSION RESULTS WITH INCOME AS DEPENDENT VARIABLE
Table C.l. Regression Results with Income as Dependent
Variable
VARIABLE COEFFICIENT STD. ERROR T-STAT. 2-TAIL SIG.
C 7399.7520 2703.9783 2.7366166 0.009
X2 187.14480 49.516736 3.7794252 0.001
X4 101.18794 38.277298 2.6435498 0.012
X5 29.387815 6.9033048 4.2570647 0.000
X6 -81.369696 28.960457 -2.8096827 0.008
X7 -228841.05 55910.163 -4.0930135 0.000
Notes: R2 =0.75 Adjusted R2 = 0.75
Some observations excluded because of missing data.
Number of observations = 45
89


APPENDIX D
REGRESSION RESULTS WITH X3 OMITTED
Table D.l. Regression Results with X3 Omitted
VARIABLE COEFFICIENT STD. ERROR T-STAT. 2-TAIL
SIG.
C -0.0381239 0.0142306 2.6790207 0.012
X2 0.0003808 0.0002629 1.4486446 0.157
X4 0.0004766 0.0001904 2.5038179 0.018
X5 5.203E-05 3.218E-05 1.6165735 0.116
X6 0.0004439 0.0001559 2.8468240 0.008
X7 -0.3204641 0.2844351 1.1266684 0.269
X8 -0.0016369 0.0016070 1.0186156 0.316
X9 0.0068546 0.0020116 3.4074334 0.002
X10 -0.0024277 0.0008096 2.9984984 0.005
Notes: R2 = 0.69 Adjusted R2 = 0.61
Some observations excluded because of missing data.
Number of observations = 40
90


APPENDIX E
DATA TABLE FOR REGRESSION ANALYSIS
Table E.l. Data Table for Regression Analysis
STATE : y XI X2 X3 X4
AL 8425 2.11 1.50 52.1
AK 0.01633 13158 14.65 0.00 64.3
AZ 0.01920 10400 5.24 0.00 54.3
AR 0.01809 8367 1.58 0.00 51.9
CA 0.01416 12668 11.54 0.93 56.6
CO 0.01372 11011 3.32 2.00 60.4
CT 0.01720 14882 6.72 0.00 61.4
DE 0.01496 11357 5.59 2.14 59.4
DC 13982 27.88 5.00 65.3
FL 0.02495 11142 4.02 0.00 54.1
GA 0.01793 9973 3.09 0.00 57.6
HI 0.02024 11264 1.63 5.27 60.6
ID 0.01315 8707 . 0.06 0.00 56.3
IL 0.00961 11787 4.37 3.30 55.6
IN 0.00718 9921 1.81 0.00 58.2
IA 0.00715 10220 0.98 4.50 58.2
KS 10675 0.19 1.70 59.6
KY 0.01291 8664 2.13 0.00 49.5
LA 8312 1.69 0.00 49.4
ME 0.00664 10119 3.78 2.61 56.0
MD 0.01213 12660 3.90 3.15 61.0
MA 0.01172 13382 12.00 2.77 58.8
MI 0.01153 10933 1.01 1.82 55.7
MN 0.00990 11243 6.05 3.45 62.1
MS 0.02646 7421 1.54 3.00 51.3
MO 0.00911 10594 2.16 0.00 57.6
MT 0.00532 9028 0.35 1.00 58.2
NE 0.00858 10443 2.97 0.00 60.5
NV 11628 0.24 0.00 65.8
NH 0.01904 12856 3.44 2.50 64.6
NJ 0.01358 14460 4.89 9.58 55.8
NM 0.01200 8463 2.94 0.00 52.8
NY 0.01040 12268 9.25 0.00 51.4
NC 0.02069 9405 4.30 2.57 59.9
ND 0.00252 9627 0.19 59.7
OH 0.01113 10421 2.65 54.0
OK 0.01854 8443 5.26 0.00 55.8
OR 0.00974 10008 2.09 0.00 58.6
PA 0.00927 10950 5.71 4.20 50.7
RI 11315 1.87 3.36 60.5
SC 0.02082 8568 3.68 3.72 56.6
91


Table E.l. (cont'd)
STATE : y XI X2 X3 X4
SD 0.00575 9520 0.00 0.00 59.8
TN 0.02029 9380 2.80 0.00 54.5
TX 0.02539 9946 2.02 0.00 58.4
UT 0.01022 8139 5.14 1.00 59.9
VT 0.00912 10170 6.22 3.20 64.0
VA 0.01164 11698 0.64 59.0
WA 11526 2.86 0.00 58.3
wv 0.00466 8023 2.84 1.50 40.4
WI 0.01072 10640 2.13 59.4
WY 0.02093 8748 3.57 2.40 59.0
Table E.l. (cont'd)
STATE X5 X6 X7 X8 X9 X10
AL 67.2 28 0.0146 0 3.16 3
AK 42.4 14 0.0240 1 3.16 1
AZ 76.2 27 0.0178 0 3.61 3
AR 39.5 37 0.0144 0 3.46 2
CA 95.7 0.0164 1 3.46 4
CO 81.7 17 0.0176 0 3.16 3
CT 92.6 15 0.0129 1 3.16 2
DE 66.0 18 0.0157 1 3.87 3
DC 100.0 0.0298 1 2.83 4
FL 90.8 25 0.0139 1 3.87 3
GA 64.6 27 0.0159 0 3.87 3
HI 76.7 12 0.0186 0 4.00 3
ID 19.6 0.0187 0 3.16 . 1
IL 82.5 0.0153 1 3.16 4
IN 68.0 21 0.0150 0 3.46 4
IA 43.1 12 0.0151 0 3.46 3
KS 52.8 17 0.0163 0 3.46 . 4
KY 45.8 30 0.0149 1 3.46 1
LA 69.0 27 0.0182 1 3.74 1
ME 36.1 16 0.0142 1 3.16 4
MD 92.9 17 0.0134 1 3.16 4
MA 90.7 14 0.0136 1 3.16 4
MI 80.2 20 0.0145 1 3.16 3
MN 66.2 9 0.0156 1 3.16 2
MS 30.3 32 0.0170 0 3.74 1
MO 66.0 21 0.0154 1 3.16 3
MT 24.2 15 0.0169 0 2.83 2
NE 47.2 11 0.0168 0 3.16 2
NV 82.6 21 0.0160 0 3.16 3
92


Table E.l. (cont'd)
STATE X5 X6 X7 X8 X9 X10
NH 56.3 14 0.0146 0 2.83 1
NJ 100.0 17 0.0128 1 3.87 3
NM 48.4 24 0.0216 0 3.87 2
NY 90.5 20 0.0141 1 2.65 1
NC 55.3 25 0.0139 0 3.87 2
ND 38.0 10 0.0194 0 2.65 3
OH 28.9 20 0.0149 1 3.46 3
OK 58.8 23 0.0164 1 3.46 2
OR 67.6 19 0.0154 1 3.16 2
PA 84.7 17 0.0134 1 3.16 3
RI 92.6 21 0.0137 1 3.87 2
SC 60.4 26 0.0149 1 3.87 2
SD 28.7 15 0.0177 0 3.16 2
TN 67.0 28 0.0145 0 3.16 1
TX 81.0 0.0183 1 4.12 3
UT 77.2 13 0.0252 0 3.87 3
VT 23.1 13 0.0146 1 3.16 2
VA 71.7 21 0.0140 0 3.16 3
WA 81.2 0.0148 1 3.16 3
WV 36.3 28 0.0137 1 3.16 3
WI 66.5 15 0.0153 1 3.16 3
WY 29.0 16 0.0185 0 3.16 2
93


NOTES
1. Department of Labor, Child Care; A Workforce Issue
([Washington D.C.]: U.S. Department of Labor, Report of
the Secretary's Task Force, 1988). All statistics cited in
the introduction were taken from this publication.
2. R. Ruopp and others, Children at the Center: Final
Report of the National Dav Care Study ed. Nancy Irwin
(Cambridge, Mass.: Abt Books, 1979).
3. Who Cares? Child Care Teachers and the Quality of Care
in America: Final Report of the National Child Care
Staffing Study. (Oakland, Calif.: Child Care Employee
Project, 1989).
4. Gary Becker, A Treatise on the Family (Cambridge,
Mass.: Harvard University Press, 1981).
5. T. W. Schultz, ed., Economics of the Family: Marriage.
Children, and Human Capital (Chicago: University of
Chicago Press, National Bureau of Economic Research, 1974) .
6. G. J. Stigler, "The Economics of Information,"
Journal of Political Economy 69 (June 1961): 213-25.
7. George A. Akerlof, "The Market for 'Lemons':
Qualitative Uncertainty in the Market Mechanism," Quarterly
Journal of Economics 84, no. 3 (August 1970): 488-500.
8. Michael Krashinsky, "Day Care and Public Policy"
(Ph.D. diss., Yale University, University Microfilms
International, 1983), 17.
9. Stigler, "The Economics of Information," 218.
10. Michael Spence, Market Signaling and Information
Transfer in Hiring and Related Screening Processes
(Cambridge, Mass.: Harvard University Press, 1974).
11. Hayne Leland, "Quacks, Lemons, and Licensing: A
Theory of Minimum Quality Standards," Journal of Political
Economy (December 1979): 1328-46.
12. It should be noted, however, that Leland adopts
Akerlof's assumption of increasing marginal opportunity
costs of supplying higher quality, whereas other
economists, notably Shapiro, assume decreasing marginal
opportunity costs with supplying higher quality services.
Leland illustrates that the primary result of under-


provision of quality is maintained in the latter case, but
that market supply is greater than optimal, rather than
less than optimal. See Leland, "Quacks, Lemons, and
Licensing."
13. Gwen Morgan, Regulation of Early Childhood Programs
(Washington, D.C.: Day Care and Child Development Council
of America, 1973), 34, citing Norris Class and Gertrude
Binder, The Licensing Responsibility in Public Welfare
(Los Angeles: [publisher not cited in secondary reference],
1957), 2.
14. Keith Leffler, "Ambiguous Changes in Product
Quality," American Economic Review 72, no. 5 (December
1982): 956-67.
15. Leffler points out, however, that it is incorrect to
assume that changes in the marginal valuation and marginal
cost of q (per unit quality) may be represented as simply
parallel shifts in the supply and demand curves. For
example, if the demand and supply curves get steeper with
quality increases the equilibrium that maximizes the summed
producer's and consumer's surplus must occur at a lower
quantity and higher price than that equilibrium wherein the
increase in marginal valuation from higher quality is
exactly equal to.the increase in marginal cost, because for
all intramarginal units the higher quality is valued more
than the increase in cost. See Leffler, "Ambiguous Changes
in Product Quality."
16. Hayne Leland, "Minimum Quality Standards and
Licensing in Markets with Asymmetric Information,"
Occupational Licensure and Regulation ed. Simon Rottenberg
(Washington D.C.: American Enterprise Institute for Public
Policy Research, 1980), 265-84.
17. In fact, licensing and similar market interventions
are sometimes justified by the argument that the consumer's
subjective expectations are consistently biased in favor of
desired outcomes and against undesired outcomes.
18. Leland, "Quacks, Lemons and Licensing," 495.
19. . Carl Shapiro, "Investment, Moral Hazard, and
Occupational Licensing," Review of Economic Studies 53
(1986): 185-98.
95