Changing Demand for Healthcare in India

Purohit BC

ISSN 0972-5997

Published Quarterly

Mangalore, India

editor.ojhas@gmail.com

Custom Search

OJHAS Vol. 18, Issue 4: (October-December 2019)

Original Article
Changing Demand for Healthcare in India

Author:
Brijesh C Purohit, Madras School of Economics, Kottur, Chennai 600025, INDIA.

Address for Correspondence
Brijesh C Purohit,
Madras School of Economics,
Kottur, Chennai 600025,
INDIA.
E-mail: brijeshpurohit@gmail.com.

Citation
Purohit BC. Changing Demand for Healthcare in India. Online J Health Allied Scs. 2019;18(4):9. Available at URL: https://www.ojhas.org/issue72/2019-4-9.html

Submitted: Dec 19, 2019; Accepted: Jan 21, 2020; Published: Feb 28, 2020

PDF

This work is licensed under a Creative Commons Attribution-No Derivative Works 2.5 India License

Abstract: Healthcare consumers may be behaving to some extent based on the nature of healthcare being a necessity or otherwise. The choice of either of public and private providers may depend upon factors like availability, accessibility, cost and quality. To some extent, this is revealed through their elasticities based on income, cost, quality and socio-economic factors. Objective of this paper is to explore the demand for healthcare services in India and estimate consumers’ elasticities to these factors. Using logit results we find that an individual may choose his preference for private or public based on distance of facility from residence. It may also depend whether it is rural or urban area and whether the income levels of state are below or above all India average. Comparing our results for two all India surveys, it is observed that people’s perception and thus preferences are mostly based on choices of nearby location, suitable timings, presence of medical personnel and less waiting time. The presence of insurance either by a national or state sponsored scheme seems to have changed the nature of healthcare demand in India from a stark necessity to a matter of better choice.
Key Words: Healthcare, availability, accessibility, quality, socio-economic factors.

Introduction:

In the healthcare sector in India, allocation of scarce fiscal resources has to be based on a criterion which meets the demand for healthcare services either from public or private providers.(1) The choice may depend upon factors like availability, accessibility, cost and quality. It is further presumed that some basic healthcare services may be among necessities. Healthcare consumers may be behaving to some extent based on the nature of healthcare being a necessity or otherwise. In this regard, their responsiveness based on income, cost, quality and socio-economic factors could be revealing. Objective of this paper is to study the demand for healthcare services in India and estimate responsiveness of healthcare consumers to these factors.

Beginning with Grossman human capital model (2,3) and its modifications, in the relevant literature, the household production function model of consumer behavior is considered. It distinguishes health as an output and medical care as one of many inputs into its production. Later elaboration of the basic Grossman approach draws a sharp distinction between fundamental objects of choice or commodities that enter the utility function and market goods and services. For example, individuals use sporting equipment and their own time to produce recreation, likewise they use medical care, nutrition, etc. to produce health. Like a firm production function, the concept of a household production function also relates specific outputs to a set of inputs. Since goods and services are inputs into the production of commodities, the demand for medical care and other health inputs is derived from the basic demand for health. Thus consumers both demand and produce health. Health is a choice variable because it is a source of utility (satisfaction) and determines income or wealth levels. Health is demanded by consumers for two reasons: i) as a consumption commodity, it directly enters their preference functions; ii) as an investment commodity, it determines the total amount of time available for market and non-market activities. An increase in the stock of health reduces the amount of time lost from these activities, and the monetary value of this reduction is an index of the return to an investment in health. Since health capital is a component of human capital, a person inherits an initial stock of health that depreciates with age, can be increased by investment, and falls below a certain level with death.

The shadow price of health is said to depend on many variables other than medical care price. Shifts in these variables alter the optimal amount of health and the derived demand for gross investment and health inputs. This price of health rises with age if the rate of depreciation on the stock of health rises over the life cycle and falls with education if more educated people are more efficient producers of health. The model stresses that, under certain conditions, an increase in the shadow price may simultaneously reduce the quantity of health demanded and increase the quantities of health inputs demanded. To develop empirically testable hypotheses, a model of the demand for health defined in terms of different indicators of mortality and diseases is specified. The model concentrates on the role of money and time prices, earned and non-earned income and health insurance. A number of socio-economic variables including religion, caste, education, assets are also used in empirical estimation. [See Annexure 1 of this paper.]

To simplify, the formal model is developed in terms of one provider of health only, but the implications for several providers can easily be drawn.

Most empirical studies use the reduced form approach and include both sets of variables denoting either demand and/or production function variables to analyze the determinants of healthcare. The conditional demand for curative care can be specified as:

[Mi|Hi=1]=b1+b2Pi+b3Vi+b4Ei+ei, i=1, 2... m sick persons

where: E is a vector of individual, household and community variables and M is the choice M=0, if taking no treatment, or taking self-treatment and other care (other than public and private) facilities; M=1, if public health facilities are used for treatment; M=2, if private healthcare is utilized. The conditional demand for curative care (above Equation) is a discrete choice model involving three choices and hence estimated using appropriate logit method.

Data Sources
In order to understand the healthcare demand in India we used data from two all India surveys namely NFHS 3 and 4. (4,5) We used two NFHS surveys, namely NFHS 3 and 4 for data. The National Family Health Survey 2015-16 (NFHS-4), the fourth in the NFHS series, provides information on population, health and nutrition for India and in each State/Union territory. NFHS-4 fieldwork for India was conducted from 20 January 2015 to 4 December 2016 by 14 Field Agencies and gathered information from 601,509 households, 699,686 women, and 112,122 men. We also compared our results from NFHS-3.

In order to understand the socio-economic correlates of healthcare demand we conducted logit analysis and also derived marginal effects or average elasticities in regard to different independent variables used in analyzing the results. The logit analysis is used here due to the fact that we are interested in looking into healthcare choices as dependent variable. We have estimated using data for all India level at four levels namely, BPL (i.e., below poverty line) rural, above BPL Rural, urban BPL and above BPL urban. This is based on all India average income of INR 77,659 (i.e. Per capita income 2015-16 at Constant 2011-12 prices) and INR 47,763 and 1,20,683 for Madhya Pradesh and Gujarat respectively. [Source:(i)Economic & Statistical Organization, Punjab (ii)Central Statistical Organization, New Delhi, 2019] Likewise, to compare between below all India average income and above all India average income level states we have estimated using data from Madhya Pradesh and Gujarat respectively (6, 7). The results of this logit analysis are discussed in the following section.

Results and Analysis

Results for BPL (Below poverty line) and above BPL rural Gujarat are presented in Tables 1 and 2 below. Only statistically significant variables are chosen for all the specifications. Thus choice of availing public or private healthcare is significantly influenced for BPL rural individual negatively by means of an individual having any health insurance (coefficient -2.102) and positively by waiting time (coefficient 2.174) (appendix 1; Table 1). It implies that an individual may choose based on his preference for private or public rather than compelled only by the nearest facility. The respective elasticities at average values of dependent variables indicate a very high responsiveness of individuals by having an insured status (marginal effect=2.005) (Table 1). In case of above BPL rural households, a dominant impact of poor quality and other reasons are notable (appendix 1; Table 2). By contrast sex (gender) variable influences the utilization of any source of healthcare. A high negative elasticity is noted for sex variable (marginal effect==-1.312, Table 2). Since sex variable captures gender, male or female, the negative coefficient may mean a choice which is more suitable to females or males based upon their preferences. Considering these results it is noteworthy that demand for availing care by rural households in Gujarat is highly responsive to insurance status, waiting time and it is dependent on whether it is male or female respondent.

Table 1: Results for BPL rural Gujarat
Number of observations =480; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Insurance status	-2.005	-1.77	0.077
Waiting	0.249	2.68	0.007
Source: estimated; note: BPL=below poverty line; ey/ex=average marginal effect; VCE = variance–covariance matrix of the estimators; OIM= observed information matrix.

Table 2: Results for above BPL rural Gujarat
Number of observations= 2,026; Model VCE : OIM; Delta-method
	ey/ex	z	P>z
Sex	-1.312	-1.77	0.077
Poor quality	0.26	2.46	0.014
Other	0.068	2.33	0.02
Source: estimated

In case of BPL urban Gujarat, no variable was found to be statistically significant and thus the results for this set of respondents are not reported here. Table 3(appendix 1) presents the logit results for above BPL urban respondents. Five variables, namely, religion, age, private insurance, poor quality and other reasons seem to be negatively influential to avail any healthcare. However, nearly responsive nature of respondents is noted for poor quality (elasticity= -.918) (Table 3). The case of Gujarat as an above average income state thus indicates that in both rural and urban areas generally respondents tend to be more responsive with gender and insurance status, waiting time and poor quality.

Table 3: Results for above BPL urban Gujarat
Number of observations=1,496; Model VCE: OIM Delta-method
Public or private care	ey/ex	z	P>z
Religion	-0.290	-2.220	0.026
Age	-0.281	-1.650	0.099
Private insurance	-0.262	-6.200	0.000
Poor quality	-0.918	-9.350	0.000
Other	-0.192	-5.210	0.000
Source: estimated

Logit results for Madhya Pradesh, a below average income state, are presented in Tables 4-7 below. It is noteworthy that choice of availing healthcare facility for BPL rural respondents is positively influenced by income and negatively influenced by no nearby facility, inconvenient timing, absent personnel and poor quality (Table 4; appendix 1). However, a high responsiveness is observed only for no nearby facility, absent personnel and poor quality (Table 4). This suggests that BPL respondents in rural areas also consider these factors seriously prior to choosing either public or private healthcare provider.

Table 4: Results for Madhya Pradesh (MP) BPL rural
Number of observations =2,383; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Income	0.073	0.760	0.450
No nearby facility	-1.526	-16.340	0.000
Inconvenient timing	-0.650	-9.190	0.000
Absent personnel	-1.049	-11.190	0.000
Poor quality	-1.701	-13.740	0.000
Source: estimated

In case of choice of availing healthcare facility by above BPL rural respondents except for insured status, which has a positive influence, all other factors, namely, income, no nearby facility, inconvenient timing, absent personnel and poor quality are depicted with a negative sign of regression coefficients for these variables. (Table 4; appendix 1). However, a high responsiveness is observed only for no nearby facility, absent personnel, waiting time and poor quality. These seem to deter going for any healthcare and thus availing healthcare from public or private care is based more on individual choice (Table 5; appendix 1). However, respondents seem to exhibit high responsiveness with regard to no nearby facility, inconvenient timing and poor quality. Thus above BPL rural respondents do exhibit that their choices could vary considerably due to various deterrents other than income or insurance (Table 5).

Table 5: Results for MP Above BPL rural
Number of observations =2,305; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Income	-0.111	-1.890	0.058
Insurance status	0.511	1.720	0.085
No nearby facility	-1.484	-16.070	0.000
Inconvenient timing	-0.929	-10.210	0.000
Absent personnel	-0.417	-6.930	0.000
Poor quality	-1.678	-15.560	0.000
Source: estimated

Unlike the rural respondents, BPL urban MP respondents seem to have only the location (i.e., no nearby facility) as a major deterrent (Table 6; appendix 1). But their responsiveness to this factor does not change the necessity of availing care (Table 6).

Table 6: Results for BPL Urban MP
Number of observations =204; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
No nearby facility	-0.350	-3.770	0.000
Source: estimated

However, for above BPL respondents in urban areas of MP, there seem to be many factors including religion, income, sex, no nearby facility, inconvenient timing, absent personnel, waiting time which discourage them to avail any type of healthcare. Only redeeming factor for these respondents is having an insurance which propels to avail necessary healthcare (Table 7; appendix 1). Yet among these factors they show more responsiveness relatively only in regard to waiting time (Table 7).

Table 7: Results for Above BPL urban MP
Number of observations =2,980; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Religion	-0.268	-3.580	0.000
Income	-0.065	-2.060	0.039
Sex	-0.322	-2.830	0.005
No nearby facility	-0.772	-14.570	0.000
Inconvenient timing	-0.492	-12.400	0.000
Absent personnel	-0.256	-7.180	0.000
waiting	-0.927	-17.770	0.000
Insurance status	0.469	3.540	0.000
Source: estimated

Results for all India Rural BPL seem to corroborate state level findings. Some of the common deterrent that emerge for this set of respondents include no nearby facility, inconvenient timing, absent personnel, waiting time, poor quality and other reasons (Table 8; appendix 1). All these variables have emerged with negative sign. The positive factors for all India Rural BPL respondents include age and sex. Nonetheless unlike State level results the all India Rural BPL do not exhibit high responsiveness pertaining to any of these factors (Table 8).

Table 8: Results for all India Rural BPL
Number of observations =96,525; Model VCE : OIM; Delta-method
Public or private care	ey/ex	z	P>z
Insurance status	0.0212	12.710	0.000
Sex	0.0321	3.610	0.000
Age	0.0478	3.090	0.002
No nearby facility	-0.755	-103.600	0.000
Inconvenient timing	-0.317	-56.180	0.000
Absent personnel	-0.112	-27.630	0.000
Waiting	-0.384	-63.230	0.000
Poor quality	-0.745	-97.450	0.000
Other reasons	-0.056	-33.310	0.000
Source: estimated

Unlike the BPL Rural all India respondents, the results of above BPL Rural respondents at all India level, indicate these factors as well as other factors like income and sex of their concern which is depicted by their negative signs (Table 9; appendix 1). Despite it their responsiveness is low for most of these factors except location or no nearby facility (Table 9).

Table 9: Results for all India Rural Above BPL
Number of observations =216,881; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Income	-0.091	-23.200	0.000
Sex	-0.057	-5.590	0.000
Insurance status	0.055	42.660	0.000
No nearby facility	-1.083	-114.510	0.000
Inconvenient timing	-0.646	-77.050	0.000
Absent personnel	-0.397	-59.450	0.000
Other reasons	-0.106	-37.900	0.000
Source: estimated

Except for insurance status, the respondents of all India Urban BPL also had deterrents such as sex, no nearby facility, personnel absent and others. (Table 10; appendix 1). In common with the rural counterparts their responsiveness is high only in regard to no nearby facility (Table 10).

Table 10: Results for all India Urban BPL
Number of observations = 5,383; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Sex	-0.054	-2.100	0.036
Insurance status	0.029	5.670	0.000
No nearby facility	-1.005	-14.780	0.000
Absent personnel	-0.225	-8.780	0.000
other reason	-0.109	-7.220	0.000
Source: estimated

Similar deterrents like income, sex, no nearby facility, inconvenient timing, personnel absent and other reasons are also depicted by the results of all India above BPL Urban respondents (Table 11; appendix 1). The difference lies in their responsiveness which is low for all of these factors ((Table 11).

Table 11: Results for all India Urban above BPL
Number of observations =140,546; Model VCE: OIM; Delta-method
Public or private care	ey/ex	z	P>z
Sex	-0.0426	-3.810	0.0000
Age	0.053	3.900	0.0000
Income	-0.212	-47.020	0.0000
Insurance status	0.049	25.370	0.0000
Employer insurance	0.001	2.220	0.0270
No nearby facility	-0.847	-92.620	0.0000
Inconvenient timing	-0.678	-70.170	0.0000
Absent personnel	-0.342	-48.810	0.0000
Other reason	0.065	-39.060	0.0000
Religion	-0.152	-20.490	0.0000
Source: estimated

If we compare the NFHS 4 results with our earlier results of NFHS 3, we find a distinct difference which is shown through an increasing responsiveness for deterrents like waiting time, inconvenient timing, absent personnel, no nearby facility which was not depicted by NFHS 3 results of ours presented in Appendix 2; Table 1a and 1b. Thus there is a change in people’s perception towards the deterrents indicating that they seem to prefer nearby location, better timings, presence of personnel, less waiting time and all this is propelled by a positive impact of having an insurance status either by the State or otherwise. Definitely thus the role of State sponsored schemes across the country seems to have changed the nature of healthcare demand from a stark necessity to a matter of better choice.

Conclusions

Our results indicate that an individual may choose based on his preference for private or public healthcare rather than compelled only by the nearest facility. The nature of results also differs whether it is rural or urban area and whether the income levels of state is below or above all India average. For instance, for above average income state we found that for above BPL rural households, a dominant responsiveness is depicted for poor quality and more suitable choices to females or males based upon their preferences. Considering our results, it is noteworthy that demand for availing care by rural households in above average state like Gujarat is highly responsive to insurance status, waiting time and it is dependent on whether it is male or female respondent.

Even for a below all India average income state like MP, our results suggest that BPL respondents in rural areas also consider some of these factors seriously prior to choosing either public or private healthcare provider. Besides in urban areas, for instance, in case of above BPL respondents of MP, there seem to be many factors including religion, income, sex, no nearby facility, inconvenient timing, absent personnel, waiting time which discourage them to avail any type of healthcare. Only redeeming factor for these respondents is having an insurance which propels to avail necessary healthcare.

Comparing our results for two surveys, namely, NFHS 4 and NFHS 3, we find that there is a change in people’s perception towards the deterrents and they seem to prefer or make better choices of nearby location, better timings, presence of personnel and less waiting time. Since in our analysis we have also included “having a health insurance” variable by any source other than private insurance, these results suggest that all this change has been propelled by a positive impact of having a health insurance either by the respective State’s sponsored or a national level scheme. Definitely thus the role of State or central sponsored schemes across the country seems to have changed the nature of healthcare demand from a stark necessity to a matter of better choice.

Appendix 1

Table 1: Results for BPL rural Gujarat
Number of observations =480; LR χ²(2) =7.41; Prob> χ² =0.0246; Log likelihood = -28.548031; Pseudo R2 =0.1149
Public or private care	Coeff.	z	P>z
Insurance status	-2.102	-1.78	0.076
Waiting	2.174	2.54	0.011
Constant	-3.069	-2.84	0.005
Source: estimated; note: coeff=coefficient; prob=probability level; LR χ²(2)=likelihood ratio chi suare

Table 2: Results for above BPL rural Gujarat
Number of observations =2,026; LR χ²(3) =10.05; Prob> χ²=0.018; Log likelihood=-78.575; Pseudo R2= 0.060
Public or private care	Coeff.	z	P>z
Sex	-1.38	-1.77	0.076
Poor quality	1.382	2.43	0.015
Other	1.821	2.25	0.025
Constant	-4.274	-5.56	0
Source: estimated

Table 3: Results for above BPL urban Gujarat
Number of observations = 1,496; LR χ²(5) = 445.60; Prob> χ² = 0.0000; Log likelihood = -726.86149; Pseudo R2 = 0.2346
Public or private care	Coef.	z	P>z
Religion	-0.468	-2.250	0.025
Age	-0.008	-1.660	0.097
Pvt insurance	-1.196	-6.970	0.000
Poor quality	-3.656	-10.040	0.000
Other	-2.72	-5.830	0.000
Constant	0.958	3.090	0.002
Source: estimated

Table 4: Results for Madhya Pradesh (MP) BPL rural
Number of observations =2,383; LR χ²(5)=2691.15; Prob> χ² =0.000; Log likelihood = -305.767; Pseudo R2= 0.814
Public or private care	Coef.	z	P>z
Income	0.206	0.750	0.451
No nearby facility	-6.281	-17.770	0.000
Inconvenient timing	-5.963	-9.750	0.000
Absent personnel	-6.267	-11.810	0.000
Poor quality	-6.860	-14.450	0.000
Constant	2.995	12.050	0.000
Source: estimated

Table 5: Results for MP Above BPL rural
Number of observations = 2,305; LR χ²(7) =2508.31; Prob> χ²= 0.0000; Log likelihood = -319.701; Pseudo R2 =0.7969
Public or private care	Coef.	z	P>z
Income	-0.443	-1.920	0.055
Insurance status	0.937	1.720	0.086
No nearby facility	-5.777	-17.540	0.000
Inconvenient timing	-5.328	-10.770	0.000
Absent personnel	-3.803	-7.320	0.000
Waiting	-5.678	-14.200	0.000
Poor quality	-5.870	-16.690	0.000
Constant	2.393	0.539	4.430
Source: estimated

Table 6: Results for BPL Urban MP
Number of observations =204; LR χ²(1)=41.30; Prob> χ² =0.000; Log likelihood = -107.01989; Pseudo R2=0.161
Public or private care	Coef.	z	P>z
No nearby facility	-2.983	-5.250	0.000
Constant	1.191	6.680	0.000
Source: estimated

Table 7: Results for Above BPL urban MP
Number of observations = 2,980; LR χ²(8) =2208.88; Prob> χ²=0.000; Log likelihood=960.362; Pseudo R2= 0.5349
Public or private care	Coef.	z	P>z
Religion	-0.625	-3.640	0.000
Income	-0.256	-2.110	0.035
Sex	-0.703	-2.860	0.004
No nearby facility	-4.385	-16.250	0.000
Inconvenient timing	-3.250	-14.090	0.000
Absent personnel	-3.306	-7.740	0.000
Waiting	-3.861	-20.500	0.000
Insurance status	1.006	3.510	0.000
Constant	2.055	5.26	0.000
Source: estimated

Table 8: Results for all India Rural BPL
Number of observations = 96,525; LR χ²(9) =75909.58; Prob> χ² = 0.0000; Log likelihood = -28155.528; Pseudo R2 =0.5741
Public or private care	Coef.	z	P>z
Insurance status	0.325	11.85	0.000
Sex	0.095	3.59	0.000
Age	0.002	3.09	0.002
No nearby facility	-3.329	-128.38	0.000
Inconvenient timing	-3.049	-63.64	0.000
Absent personnel	-1.906	-29.9	0.000
Waiting	-2.722	-72.19	0.000
Poor quality	-3.362	-116.37	0.000
Other reason	-2.853	-46.75	0.000
Cons	2.389	51.5	0.000
Source: estimated

Table 9: Results for all India Rural Above BPL
Number of observations =216,881; LR χ²(7)= 147679.82; Prob> χ²=0.0000; Log likelihood =-75116.1; Pseudo R2=0.4957
Public or private care	Coef.	z	P>z
Income	-0.324	-23.880	0.000
Sex	-0.139	-5.600	0.000
Insurance status	0.587	37.690	0.000
No nearby facility	-5.528	-121.740	0.000
Inconvenient timing	-5.457	-80.630	0.000
Absent personnel	-5.189	-62.110	0.000
Other reasons	-5.437	-40.590	0.000
Cons	1.628	62.890	0.000
Source: estimated

Table 10: Results for all India Urban BPL
Number of observations =5,383; LR χ²(5) =2518.22; Prob> χ² = 0.0000; Log likelihood = -2386.577; Pseudo R2= 0.3454
Public or private care	Coef.	z	P>z
Sex	-0.176	-2.130	0.033
Insurance status	0.452	4.850	0.000
No nearby facility	-5.885	-15.430	0.000
Absent personnel	-4.789	-9.430	0.000
Other reason	-4.698	-8.010	0.000
Constant	1.165	15.820	0.000
Source: estimated

Table 11: Results for all India Urban above BPL
Number of observations =140,546; LR χ²(10) =80195.06; Prob> χ² =0.000; Log likelihood = -57242.396; Pseudo R2 =0.4119
Public or private care	Coef.	z	P>z
Sex	-0.091	-3.820	0.000
Age	0.002	3.890	0.000
Income	-0.779	-51.340	0.000
Insurance status	0.401	23.380	0.000
Employer insurance	0.179	2.140	0.032
No nearby facility	-4.546	-100.470	0.000
Inconvenient timing	-4.685	-74.220	0.000
Absent personnel	-4.323	-51.500	0.000
Other reason	-3.087	-49.370	0.000
Religion	-0.367	-20.930	0.000
Constant	1.451	38.720	0.000
Source: estimated

Appendix 2

Results from NFHS3 data

Table 1a. Rural all India results for Public Private Care
ANYCARE	Coeff.	z	P>z	ey/ex
NONFACTY	-0.212	-11.340	0.000	-0.0089
TIMENC	-0.4394	-15.150	0.000	-0.0106
HPABST	1.2374	31.520	0.000	0.0288
WAITTL	-0.0119	-0.470	0.637	-0.0003
PQUAC	-0.9981	-55.150	0.000	-0.0448
BPL	0.032	2.490	0.013	0.0015
INSANY	0.0696	8.450	0.000	0.0025
CASTE	-0.0476	-5.900	0.000	-0.011
WATSS	0.028	27.250	0.000	0.0608
SANTYP	0.0178	14.750	0.000	0.0438
WI	-0.2355	-22.350	0.000	-0.0523
RELGN	0.0004	0.320	0.748	0.0001
FEEDU	-0.0433	-6.130	0.000	-0.0035
ELECTR	-0.0787	-3.440	0.001	-0.0055
Constant	2.3557	40.650	0.000	-
Source: (Purohit, 2013;) estimated. ANYCARE, any type of healthcare; Coeff., coefficient; NONFACTY, no-nearby facility; TIMENC, facility timing not convenient; HPABST, health personnel often absent; WAITTL, waiting time too long; PQUAC, poor quality of care as perceived by the respondents; BPL, BPL card holding; INSANY, insurance coverage from any source; CASTE, caste; WATSS, source of water supply; SANTYP, type of sanitation; WI, wealth index; RELGN, religion; FEEDU, female education; ELECTR, having electricity. Number of observations=147743; logistic regression χ²(14)=8587.59; Prob>χ²=0.0000; log likelihood=-45884.029; pseudo R2=0.0856.

Table 1b: Urban all India results for Public or Private Care
ANYCARE	Coeff.	z	P>z	ey/ex
NONFACTY	-0.4703	-26.890	0.000	-0.0348
TIMENC	-0.3613	-14.630	0.000	-0.0159
HPABST	0.6106	16.050	0.000	0.021
WAITTL	-0.2132	-11.020	0.000	-0.0134
PQUAC	-0.5465	-31.290	0.000	-0.0457
BPL	-0.0169	-1.350	0.177	-0.001
INSANY	0.0226	3.580	0.000	0.0018
CASTE	0.0335	4.180	0.000	0.0159
WATSS	0.0039	4.640	0.000	0.012
SANTYP	0.0151	10.560	0.000	0.0427
WI	-0.2536	-18.830	0.000	-0.1658
RELGN	-0.0023	-1.240	0.213	-0.0006
FEEDU	-0.095	-16.410	0.000	-0.0296
ELECTR	1.0537	30.700	0.000	0.1818
Constant	1.86	28.800	0.000	-
Source: (Purohit, 2013); estimated. ANYCARE, any type of healthcare; Coeff., coefficient; NONFACTY, no-nearby facility; TIMENC, facility timing not convenient; HPABST, health personnel often absent; WAITTL, waiting time too long; PQUAC, poor quality of care as perceived by the respondents; BPL, BPL card holding; INSANY, insurance coverage from any source; CASTE, caste; WATSS, source of water supply; SANTYP, type of sanitation; WI, wealth index; RELGN, religion; FEEDU, female education; ELECTR, having electricity. Number of observations=98284; logistic regression χ²(14)=4576.65; Prob>χ²=0.0000; log likelihood=-43386.578; pseudo R2=0.0501.

Annexure 1:

Following the general tradition in the literature (8 to 15), we assume that individual “x” in a given period faces ‘Y’ healthcare provider alternatives.( For each “y” alternative, the individual’s utility is given by the conditional utility function:
Uxy=U(Hx, y, Cx,y) (1)
Where Hxy=expected health status of individual “x” after receiving care from provider “y”; Cxy==consumption of goods other than healthcare, when an individual “x” chooses healthcare provider “y”.
A simple budget constraint is defined as:
Ix= Cxy+TPxy (2)
Where Ix=individual’s income and TPxy is the total price of choosing provider “y”. The total price is formed by two components: monetary and non-monetary price. Then
Ix=Cxy+ (Py +Txy) (3)
Where Py represents the monetary price of provider “y”(which is identical for all individuals; price discrimination is not allowed) and Txy is the non-monetary price which is measured as the opportunity cost of time devoted to travelling and waiting in the “y” provider’s choice.
Let TTxy and WTxy represent travel time and waiting time associated with the choice of alternative “y” and Wx be the opportunity cost of time for individual “x”, then:
Txy=(wx) * (TTxy+wTxy) (4)
Provider’s price affects the contact decision as different proportion of the individual’s income remains available for consumption of other goods.
Expected health status after being treated by provider “y”is represented by two additive factors: the expected health status with alternative 0, with y=0 being the case of self-care in the absence of formal treatment by a healthcare provider; and the expected effectiveness of another alternative “y ” in relation to y=0. That is Hxy=Exy +Hx0. (5)
Where Exy=expected effectiveness (or quality measure) of provider “y” and Hx0=expected health status from the choice of provider 0. Then expected effectiveness may be represented as a household production function which depends on patient and provider characteristics.
Exy=E(Patx, Zy) (6)
Where Patx is a vector of individual patient characteristics whose effect varies between alternatives (effectiveness and service quality perceived by the individual), and Zy is a vector of provider characteristics.
The conditional utility function may now be expressed by substituting (3), (4) and (5) into (1):
Uxy=U(Hx0+Exy, Ix-Py-wx * TTxy- w * WTxy) (7)
If Uxh being the highest utility that an individual may obtain, the unconditional utility maximization problem for the individual “x” in period “t” takes the form:
Uth= max (Ux0, Ux1, Ux2 Uxy) (8)
It is presumed that the consumer chooses a provider based upon the utility maximization or perceived effectiveness of treatment through a particular provider and thus in the empirical specification we use a multinomial logit model which considers a choice among different alternative providers and thus their demand (or demand for a particular kind of healthcare public or private) is estimated using an equation of the following general form:
Demand or Choice of a particular provider or treatment Yd=F (Patx, Zy, Py, Txy) +exy
where “exy”is a zero mean random disturbance term with finite variance and is uncorrelated across individuals.

References

Purohit BC. Demand for healthcare in India. Healthcare in Low-resource Settings 2013; 1:e7, pp. 25-30; doi:10.4081/hls.2013.e7; PAGEPress, Italy.
Grossman M. On the concept of health capital and the demand for health. J Pol Econ. 1972;80: 223–55.
Grossman M. Front matter, the demand for health: a theoretical and empirical investigation.In The demand for health: a theoretical and empirical investigation 1972 Jan 1 pp. 20-0), NBER.
International Institute of Population Sciences, National Family Health Survey, 2015-16, Government of India.
International Institute for Population Sciences, National Family Health Survey (NFHS-3), India, 2005-06: state level reports. Mumbai: IIPS ed.; 2008
Government of Punjab, Economic & Statistical Organization, Punjab
Government of India, Central Statistical Organization, New Delhi,2019
Ii M. The demand for medical care evidence from urban areas in Bolivia. The living standard measurement study”, working paper no.123. Washington, DC: The World Bank; 1993.
Mwachofi A, Al-Assaf F. Health care market deviation from the ideal market. Sultan Qaboos University Medical Journal. 2011.
Wellay et al. Demand for health care service and associated factors among patients in the community of Tsegedie District, Northern Ethiopia. BMC Health Services Research 2018;18:697. https://doi.org/10.1186/s12913-018-3490-2.
Puig-Junoy J, Saez M, Martinez-Garcia E. Health care provider choice in the case of patient initiated contacts: an extended version of discrete choice model demand. Available at https://www.researchgate.net/profile/Esther_Martinez-Garcia/publication/37762010_Health_Care_Provider_Choice_in_the_Case_of_Patient-Initiated_Contacts_An_Extended_Version_of_Discrete_Choice_of_Model_Demand/links/0c960529dc4291a8de000000/Health-Care-Provider-Choice-in-the-Case-of-Patient-Initiated-Contacts-An-Extended-Version-of-Discrete-Choice-of-Model-Demand.pdf.
Rosenzweig MR, Schultz TP. Estimating a household production function: heterogeneity,the demand for health inputs and their effects on birth weight. J Polit Econ 1983;91:723-46.
Schultz T.P. Studying the impact of household economic and community variables on child mortality. Popul Dev Rev 1984;10: S25-45.
Behrman JR, Deolalikar A. B. Health and nutrition. In: Chenery H, Srinivasan TN, eds. Handbook of development economics. Amsterdam: North Holland Press; 1988, pp 631-771.
Strauss J, Thomas D. Health, nutrition and economic development.J Econ Lit 1998;36:766-817.
Babalola O. Consumers and Their Demand for Healthcare. J Health Med Econ. 2017;3(1):6.