Paying for the view? How nursing home prices affect certified staffing ratios

Heger, Dörte; Herr, Annika; Mensen, Anne Article — Published Version Paying for the view? How nursing home prices affect certified staffing ratios Health Economics Provided in Cooperation with: John Wiley & Sons Suggested Citation: Heger, Dörte; Herr, Annika; Mensen, Anne (2022) : Paying for the view? How nursing home prices affect certified staffing ratios, Health Economics, ISSN 1099-1050, Wiley, Hoboken, NJ, Vol. 31, Iss. 8, pp. 1618-1632, https://doi.org/10.1002/hec.4532 This Version is available at: https://hdl.handle.net/10419/265078 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ 1618 1 | INTRODUCTION Demographic change and population aging will lead to a steep increase in the demand for long-term care (LTC) over the next decades. Additionally, informal care giving is likely to decline due to the increased labor force participation of women, which boosts demand for LTC workers even further (Colombo etal.,2011). Consequently, LTC systems worldwide face the challenge of how to deliver and finance high-quality LTC services (Spasova etal.,2018). This paper highlights the case of Switzerland, one of the richest Organisation for Economic Co-operation and Development (OECD) countries with a gross domestic product (GDP) per capita of more than US$73,000 (OECD,2021a). Spending on LTC amounts to 2.4% of the GDP (OECD,2021b). Moreover, Switzerland is among the OECD countries with the highest supply of, as well as demand for, nursing home care. In 2016, 153,301 patients were cared for in 1552 nursing homes, which is equivalent to 10.1% of the Swiss population aged 65years or older (Federal Office of Public Health (FOPH)(2021); Federal Statistical Office,2018). Swiss nursing homes have high nurse staffing ratios (0.5 full-time equivalent nurses per resident) compared to, for example, the US with 0.4 (Dyer 1RWI – Leibniz Institute for Economic Research, Essen, Germany 2Leibniz Science Campus Ruhr, Essen, Germany 3Institute of Health Economics and CHERH, Leibniz University Hannover, Hannover, Germany 4CINCH – Health Economics Research Center, Essen, Germany 5Ruhr-University Bochum, Bochum, Germany Correspondence Anne Mensen, RWI – Leibniz Institute for Economic Research, Hohenzollernstraße 1-3, 45128 Essen, Germany. Email: [email protected] Funding information Federal Ministry of Education and Research, Grant/Award Number: 01EH1602B; Leibniz-Gemeinschaft Open access funding enabled and organized by Projekt DEAL. Abstract Many countries limit public and private reimbursement for nursing care costs for social or financial reasons. Still, quality varies across nursing homes. We explore the causal link between case-mix adjusted nurse staffing ratios as an indicator of care quality and different price components in Swiss nursing homes. The Swiss reimbursement system limits and subsidizes the care price at the cantonal level, which implicitly limits staffing ratios, while the residents cover the nursing home-specific lodging price privately. To estimate causal effects, we exploit (i) the exogeneity of the Swiss care price regulation, (ii) nursing-home fixed effects estimations and (iii) instrumental variables for the lodging price. Our estimates show a positive impact of prices on certified staffing ratios. We find that a 10% increase in care prices increases certified staffing ratios by 3–4%. A comparable 10% increase in lodging prices raises certified staffing ratios by 1.5–10% (depending on the model). Our findings highlight that price limits for nursing care impose a limit on staffing ratios. Furthermore, our results indicate that providers circumvent price limits by increasing lodging prices that are privately covered. Thus, this cost shifting implicitly shifts the financial burden to the residents. KEYWORDS care quality, long-term care, nursing home, prices, staffing ratios RESEARCH ARTICLE Paying for the view? How nursing home prices affect certified staffing ratios Dörte Heger1,2 | Annika Herr3,4 | Anne Mensen1,2,5 DOI: 10.1002/hec.4532 Received: 4 January 2021 Revised: 4 March 2022 Accepted: 25 April 2022 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2022 The Authors. Health Economics published by John Wiley & Sons Ltd. wileyonlinelibrary.com/journal/hecHealth Economics. 2022;31:1618–1632. 1619 etal.,2019). Since LTC is highly labor-intensive and the population is aging rapidly, the demand for LTC workers will increase steadily in most OECD countries (Colombo etal.,2011). As elsewhere, the Swiss system is struggling to address rising costs and expected staff shortages (Cosandey & Kienast,2016). To keep rising costs under control, the Swiss parliament passed a law in 2008 to implement a new financing system that became effective in January 2011 and regulates payments of the obligatory health insurance including capped private contributions for nursing care at a fixed level across cantons (The Federal Council,2021b). In this paper, we address the interrelation between financing nursing home care, certified staffing levels, and public-private cost-sharing. We analyze (i) whether higher prices lead to better care quality, and (ii) whether price limits on nursing care increase privately borne nursing home costs. Since certified staffing ratios are shown to be a good indicator of care quality (Campbell etal.,2000; Konetzka etal.,2008; Lin,2014), higher prices may increase care quality thanks to higher nurse staffing ratios. We follow Hackmann(2019) and Lin(2014) by measuring care quality by case-mix weighted certified nurse staffing ratios, that is, the number of certified nurses divided by the number of residents. To study the relationship between prices and certified staffing ratios, we exploit a unique feature of the Swiss LTC system that allows us to draw causal inferences. The nursing home providers are reimbursed separately for nursing care and for accommodation and services. In the 12 cantons we analyze, providers face canton-wide fixed prices for nursing care (care price), while they set individual prices for accommodation and services (lodging price). This canton-specific regulation enables us to analyze (i) the causal effect of care prices on certified staffing ratios, and (ii) potential cost-shifting to the residents by investigating the extent to which lodging prices affect certified staffing ratios. Residents fully cover the lodging price, but only pay a capped contribution to nursing care. Hence, subsidizing public with private payments would raise equity concerns. Anecdotal evidence indicates that for some nursing homes lodging prices lie above the actual costs for lodging, thus, they may use the excess amount to finance care costs (Cosandey & Kienast,2016; Federal Statistical Office,2015). Newspaper articles suggest that around 50% of nursing homes substitute their care costs with reimbursement for lodging (Erny & Weber,2018 (SRF)). When explaining the relation between the lodging price and nurse-to-resident ratios, we deal with potential endogeneity concerns in our linear regressions, for example, when both are driven by unobserved demand characteristics. We follow two approaches. First, we present regressions including nursing home fixed effects that control for time-invariant unobservable characteristics. Second, we draw from Forder and Allan(2014) as well as Herr and Hottenrott(2016) and apply a two-stage least squares approach and construct an instrumental variable for the lodging price. The direction of the effect of prices on certified staffing ratios is a priori uncertain. On the one hand, higher prices may increase quality, as it is possible to employ more registered nurses. On the other hand, nursing homes can be classified as experience goods and higher prices may be used to signal high quality to potential residents irrespective of the actual quality level (Plassmann etal.,2008), especially in the Swiss case, where care prices are capped and lodging prices should not be used to increase care staffing. The results are also relevant for other countries with financially strained LTC systems. For example, German nursing homes are also required to set separate prices for nursing care and for accommodation and services, while this separation is only enforced ex-ante. However, the issues of implicit quality limits and equity concerns remain. In contrast, US nursing homes only charge a single price to cover all costs and may set prices for private payers freely, while public Medicare and Medicaid contributions are tightly regulated with different regulations by state and population group (Feder etal.,2000). Hence, while regulations may differ across countries, the aim of the regulation is always to restrict public or private expenditures for nursing homes. Fixed care prices lead to the question of how nursing homes cover higher costs if they want to employ more registered nurses. From a competition theory point of view, a fixed price combined with unobservable differences in quality may lead to a convergence in the quality across nursing homes. If asymmetric information is relevant, this convergence may result in decreased certified staffing ratios, even lower than the residents would be willing to pay for. This adverse selection problem was first identified in the market for used cars by Akerlof(1970), but adverse selection also occurs in other markets such as, for example, the market for health or life insurance (Dionne etal.,2000) or financial markets (Philippon & Skreta,2012). To prevent adverse selection, nursing homes may use higher lodging prices either as a signal or as an actual subsidy to deliver higher quality. Although such cross-subsidization is unlawful, it is frequently reported (INFRAS etal.,2018). While price regulation is effective in limiting public expenditures, the issue of securing a high quality of care by maintaining an adequate level of certified staffing remains (Spasova etal.,2018). Reports about insufficient care in nursing homes are common (e.g., Tscharnke(2009)). A central aspect of such reports and in scientific debates on care quality are nurse-to-resident staffing ratios (see e.g., Chen and Grabowski(2015) or Tong(2011), on US nursing homes). Although there are several attempts to measure outcome quality objectively, this remains difficult in the LTC context due to a lack of objective and comparable data (Castle & Ferguson,2010). HEGER Et al. 1620 The literature mostly points to a positive relationship between care prices and care quality. Looking at Medicaid reimbursement rates in the US both Cohen and Spector(1996) and Grabowski(2001) find a positive impact of prices on staffing ratios. Similarly, a recent study by Hackmann(2019), which uses a structural model of the nursing home industry in the US state of Pennsylvania, shows that a 10% increase in the Medicaid reimbursement rate leads to an 8.7% increase in registered nurse staffing ratios. Moreover, it shows that this would also increase overall welfare, despite the higher costs. In the German context, Reichert and Stroka(2018) find that some medical quality indicators are positively correlated with prices while others do not show any significant relation. Like-wise, Forder and Allan(2014) find that competition reduces prices, which in turn pushes down quality in English nursing homes. Herr and Hottenrott(2016) provide causal evidence on the direct link between higher prices and better quality of care in German nursing homes. Our study contributes to the literature in two ways. First, it exploits the unique Swiss reimbursement system to measure the impact of care prices on quality of care and also addresses endogeneity concerns present in previous studies. Second, it uses case-mix weighted certified staffing ratios to measure care quality, which allows comparability between findings from different countries. Our results show that a 10% increase in care prices increases the number of registered nurses per resident by 3.4–4%. Furthermore, a 10% increase in the lodging price raises the certified staffing ratio by 1.5% in the fixed-effects model and by 10% in the two-stage least squares estimation, which is in line with the cross-subsidization theory. Hence, higher lodging prices may serve as a signal of higher quality since care prices do not vary within a canton. 2 | METHODS 2.1 | Literature on quality of care and certified nurse staffing ratios The measurement of quality of care remains a debated topic. Donabedian's model classifies quality of care by distinguishing between structural, process, and outcome parameters (Donabedian,1988). Structural parameters refer to organizational factors that define the health care system. Nurse staffing is an important structural parameter for care quality that has been shown to have a direct and indirect impact on different care outcomes (Campbell etal.,2000; Konetzka etal.,2008; Lin,2014). Among others, these are improved physical functioning, less antibiotic use, fewer deficiencies, less weight loss, less dehydration, and even fewer hospital visits and lower mortality rates (Friedrich & Hackmann,2021; Kaiser Family Foundation (KFF),2018; Kimmey & Stearns,2015; Lin,2014). Nevertheless, nationwide nurse staffing standards are lacking in many countries and existing standards are difficult to compare due to differences in measurement and the vagueness of standards (Harrington etal.,2012). Swiss cantons implemented different staffing requirements or minimum educational requirements for LTC workers. For example, minimum staffing ratios exist for all nursing homes and most cantons define quantitative as well as qualitative minimum staffing ratios that require a specific share of qualified personnel. However, staffing requirements are not always binding in the short run and adherence is not publicly reported (Cosandey & Kienast,2016). Furthermore, the specified staffing requirements are typically considered to be insufficient and to provide a lower limit for quality (Harrington etal.,2012; Mueller et al., 2006). Indeed, inadequate staffing ratios are an important reason for quality shortfalls in nursing homes (Harrington etal.,2016, 2020). We follow Hackmann(2019) and Lin(2014) and measure care quality by case-mix weighted certified staffing ratios. The case-mix weight is necessary since individuals with a higher care level require more care than individuals with a lower care level. This is especially the case for qualified staff: A recent study documenting the workflow and measuring the necessary time and qualification requirements in nursing homes in Germany shows that the optimal certified nurse-to-resident level increases continuously with the level of care needed while the pattern is less clear for nursing assistants (Rothgang,2020). 2.2 | Institutional setting The organization of LTC in Switzerland largely falls under the responsibility of the 26 cantons, which enter contracts with nursing homes regarding the amount of care provided, reimbursement, and operating conditions directly or they delegate this duty to the municipalities. In general, reimbursement for nursing homes consists of three components: The care price, the price for accommodation and services, and subsidies. However, the amount of reimbursement of nursing care costs varies across Swiss cantons and for some cantons also across nursing homes. While many cantons set predetermined prices for the reimbursement of nursing care that apply to all nursing homes within the canton – so-called “norm costs” – or define HEGER Et al. 1621 canton-wide maximum reimbursement levels, some cantons have community-based or provider-specific regulations that vary across nursing homes within a canton (for an overview of the financing schemes, see TableA1 in the appendix). In this study, we only include nursing homes from the 14 cantons with fixed prices or maximum reimbursement rates. Among these cantons, only two cantons define a maximum reimbursement per level of care, which we exclude in a robustness check. In addition, in each canton the care price depends on the resident's intensity of care. Thus, in our sample, the care price is exogenous for the single nursing homes and only varies by the resident's care level. Upon entering a nursing home, a patient's LTC needs are assessed on a scale of 0–12 care levels based on a detailed clarification of needs process where each care level represents 20min of daily care needs. The patient or their relatives and the nursing staff answer questions on the patient's physical and cognitive abilities and limitations, well-being, nutrition, continence, pain, plus medication and therapies. The needs assessment is repeated regularly to (re-)assess the patient's level of care (further information about the process is provided in AppendixA1). The financing of nursing care is split between the obligatory health insurance, the dependent person, and the cantons. Based on the assigned care level, the insurance pays a contribution to the nursing home for the provision of nursing care of 9 CHF (1 CHF equals approximately US $1.1 on June 21, 2021) per day and level of care, that is, the insurance contribution is capped at 108 CHF per day for the 12th care level (9 CHF x 12). The private contributions to nursing care are limited at the federal level and may not exceed 20% of the maximal reimbursement by the insurance. Hence, the patient's contribution is capped at 21.60 CHF (108 CHF x 0.2) per day for all cantons. Any remaining nursing care costs up to the norm costs or maximum reimbursement levels are financed by the cantons (Cosandey & Kienast,2016). Table1 gives an example of how the care price is split up. The lodging price is set freely by each nursing home and is borne completely by the resident. Nursing homes also receive subsidies, for example, to cover losses or to finance new buildings or equipment. Subsidies represent a minor part of the total reimbursement and are paid by the municipalities, the cantons, or foundations. While the regulation of nursing home prices is effective in limiting the cantonal expenditures for nursing home care, the fixed reimbursements for nursing care leave little room for competition based on care quality since the reimbursement levels are targeted at covering costs based on past cost accounting. Hence, nursing homes may have an incentive to use lodging prices to increase staffing ratios. Besides prices (certified) staffing ratios are the only other quality-related measures that are both comparable and available (Cosandey & Kienast,2016). 1 Thus, we assume that nursing homes use additional revenue through cross-subsidization to hire additional certified staff and signal higher quality. Although subsidizing nursing care through higher lodging prices is illegal, it is expected to be common practice in Switzerland (INFRAS etal.,2018). Cross-subsidization, however, raises equity concerns since this price component is fully borne by the resident. HEGER Et al. Level of care Norm costs Health insurance Resident Canton/municipality 1 14.80 9.00 5.80 0.00 2 38.10 18.00 20.10 0.00 3 48.60 27.00 21.60 0.00 4 70.30 36.00 21.60 12.70 5 97.90 45.00 21.60 31.30 6 115.60 54.00 21.60 40.00 7 137.00 63.00 21.60 52.40 8 150.00 72.00 21.60 56.40 9 175.70 81.00 21.60 73.10 10 183.10 90.00 21.60 71.50 11 206.40 99.00 21.60 85.80 12 277.40 108.00 21.60 147.80 Note: Prices are given in Swiss francs per care day. The norm costs represent the canton-wide fixed care price and vary by canton and year (2012–2017). The part paid by the health insurance is fixed for all cantons across Switzerland and the amount paid by the resident is capped at 21.60 CHF. The norm costs shown here serve as an example (canton Thurgau). The cantons may delegate the financing obligations to the municipalities. Source: Alters-und Pflegeheim National(2013) showing taxes for the year 2013. TABLE 1 Example of financing mix for nursing care (canton Thurgau in 2013) 1622 2.3 | Data We base the analysis on administrative data published yearly by the Swiss Federal Office of Public Health including key performance figures and characteristics of all Swiss nursing homes from 2012 to 2017 (FOPH,2021). 2 Our dependent variable is the case-mix weighted number of registered nurses per 1000 care days (certified staffing ratios). We account for the residents' average case-mix to control for the fact that the medical needs and skilled nursing needs increase with the resident's care level. Since the data comprises the average care level of each nursing home, we define the case-mix weight as follows: We divide the average cantonal care level by the average care level of the nursing home within this canton. This accounts for the fact that nursing homes with a care level below the canton average require fewer registered nurses to provide the same quality level and vice versa. Certified staffing ratios of nursing home i in canton c and year t are given by: Certified staffing ratio 𝑖𝑡 = #of registered nurses 𝑖𝑡 1,000 care days ⋅ average care level𝑐𝑡 care level𝑖𝑡 As explanatory variables of interest, we analyze the different price components. We consider the lodging price, the care price, and revenues from subsidies, where subsidies may involve all kinds of financial support from the public sector or private organizations. We compute revenues for accommodation and services by the residual between the two former revenue components and total revenues. Analogous to the certified staffing ratio, we weight the care price by the nursing home's case-mix. All prices are measured as prices per care day in Swiss Francs. We control for nursing home size by using the number of beds including long-term, short-term, and acute care beds. Moreover, we use an indicator of whether a nursing home is a private for-profit home to allow for potential correlation between the certified staffing ratio and ownership type. Comondore etal.(2009) find in their meta study that, on average, non-profit nursing homes provide higher quality than for-profit nursing homes. Furthermore, Grabowski etal.(2013) identify a positive causal effect of non-profit ownership on quality using instrumental variable estimations. 3 Here, state-run and non-profit nursing homes serve as the reference group.In addition, we control for the cantonal price level (GDP per capita), the cantonal age structure (share of population aged 65 and above) and the cantonal unemployment rate to account for changes in buying power and needs. Since our estimation strategy relies on exogenously set nursing prices, we select the cantons that have limited their care prices by defining fixed norm costs (10 cantons) or by setting an upper bound (2 cantons; Cosandey & Kienast,2016; INFRAS etal.,2018). Thus, care prices by level of care intensity are predetermined for all nursing homes in our sample. For our regression analysis, we exclude the two very small cantons with less than 10 nursing homes. This results in 5668 nursing home-year observations. Of these, we exclude 34 observations without any beds for LTC. To further reduce the potential influence of outliers, we only keep observations where our main variables (number of registered nurses, price components, number of beds) lie within the first and the 99th percentile excluding 367 more observations. Lastly, since the construction of our instrumental variable relies on the existence of similar nursing homes, we only include nursing homes with comparable competitors within a specified region (details below). This leads to our final sample of 4390 observations from 925 nursing homes over 6 years. 2.4 | Descriptive statistics Table2 shows the average values and standard deviations of our main variables for the whole sample as well as by canton. The average nursing home has around 51 beds and private for-profit owners run 33% of all nursing homes. Swiss nursing homes employ, on average, slightly less than one registered nurse per 1000 care days or 0.35 (0.97/1000 x 365) registered nurses per patient. All numbers denote case-mix weighted full-time equivalents. The average number of registered nurses per 1000 care days varies from 0.67 in Aargau to 1.48 in Jura with standard deviations between 0.21 and 0.35, implying considerable quality differences across and within cantons. The average cantonal care price varies between 84.13 and 126.33 CHF per care day. In comparison, the average lodging price varies between 125.85 and 205.31 CHF across cantons and resulting revenues account for the largest part of the total reimbursements in all cantons. Subsidies only account for a minor part of revenues (1.42 to 8.82 CHF per care day, on average). Figure1 shows the variation in prices and certified staffing ratios at the nursing-home level over time. Specifically, it displays the cantonal averages of the yearly variation in certified staffing ratios and the price components within nursing homes, HEGER Et al. 1623HEGER Et al. All AG AR BE BL BS GR JU SG SO TG VD VS Registered nurses per 1000 care days 0.97 (0.34) 1.04 (0.33) 0.67 (0.22) 1.09 (0.34) 1.06 (0.23) 1.22 (0.29) 1.01 (0.35) 1.48 (0.27) 0.74 (0.21) 1.07 (0.35) 0.91 (0.22) 0.74 (0.29) 0.94 (0.23) Daily prices (in CHF) Nursing care 107.71 (17.87) 95.83 (10.61) 84.13 (11.38) 113.43 (16.82) 97.25 (5.79) 109.09 (12.78) 106.21 (12.20) 115.42 (14.42) 89.43 (7.82) 104.06 (13.84) 98.86 (9.31) 122.42 (15.37) 126.33 (3.64) Lodging 157.09 (30.30) 162.45 (29.37) 125.85 (32.72) 162.76 (27.86) 205.31 (14.33) 193.21 (39.86) 146.45 (26.23) 151.41 (14.56) 144.48 (24.78) 162.97 (29.22) 146.91 (22.32) 160.02 (29.48) 127.80 (13.42) Subsidies 4.66 (12.62) 2.65 (11.38) 8.54 (17.94) 3.22 (11.17) 1.42 (3.65) 7.02 (20.60) 9.09 (19.20) 9.82 (12.02) 4.43 (10.95) 5.57 (17.59) 2.10 (6.97) 7.30 (11.99) 6.54 (6.29) Number of beds 50.94 (29.96) 56.11 (31.02) 32.00 (12.18) 45.25 (30.50) 105.76 (27.45) 75.02 (27.74) 47.21 (22.54) 48.89 (8.37) 55.28 (27.22) 56.83 (21.63) 48.37 (25.51) 47.15 (28.99) 68.69 (33.75) Private nursing homes (in %) 33.21 (47.10) 33.40 (47.21) 43.02 (49.80) 44.16 (49.67) 0.00 (0.00) 4.76 (21.38) 4.40 (20.55) 0.00 (0.00) 16.89 (37.50) 25.99 (43.96) 61.00 (48.90) 43.29 (49.59) 1.03 (10.13) Population 65+ (in %) 18.47 (1.62) 16.76 (0.40) 18.65 (0.37) 20.01 (0.37) 20.97 (0.40) 20.29 (0.23) 19.84 (0.73) 19.68 (0.62) 17.33 (0.44) 18.81 (0.36) 16.57 (0.47) 16.23 (0.12) 18.56 (0.53) Unemployment Rate (in %) 2.87 (0.92) 2.99 (0.17) 1.76 (0.13) 2.45 (0.23) 2.77 (0.10) 3.74 (0.10) 1.67 (0.08) 3.95 (0.56) 2.39 (0.08) 2.67 (0.18) 2.41 (0.14) 4.79 (0.18) 3.93 (0.30) GDP per capita (in 1.000 CHF) 75.19 (19.60) 64.31 (0.19) 57.83 (1.00) 77.98 (0.58) 71.17 (0.86) 180.73 (9.93) 72.16 (0.74) 65.38 (1.31) 74.84 (0.58) 67.55 (0.42) 62.04 (0.39) 72.00 (1.47) 54.46 (0.15) Urbanity 0.09 (0.28) 0.00 (0.00) 0.00 (0.00) 0.11 (0.32) 0.00 (0.00) 0.99 (0.09) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.14 (0.34) 0.00 (0.00) Number of nursing homes per year 731.83 80.04 14.44 265.51 5.62 21.29 41.67 3.42 98.70 38.06 33.36 99.37 32.40 Observations 4,390 479 86 1592 29 126 250 19 592 227 200 596 194 Note : The number of registered nurses and care prices are weighted by case-mix. Means and standard deviations (in brackets) are generated from the pooled cross-section (years 2012–2017). Cantons: AG=Aargau, AR: Appenzell Ausserrhoden, BE: Bern, BL: Basel-Landschaft, BS: Basel-Stadt, GR: Grisons, JU: Jura, SG: St. Gallen, SO: Solothurn, TG: Thurgau, VD: Vaud, VS: Valais. Source: FOPH2021, 2012–2017, own calculations. TABLE 2 Descriptive statistics by canton (mean values, SD given in parentheses) 1624 where the size of the circles reflects the underlying number of nursing homes in the canton. Care and lodging prices increase by 2.67% per year, while staffing ratios increase, on average, by 4.49% per year. 2.5 | Estimation strategy The empirical strategy exploits the geographical and time variation of care prices, that is, our identifying variation stems from price variation across cantons over time. As a first step, we use ordinary least squares (OLS) estimation to regress certified staffing ratios in nursing home i in year t on the different price components and other nursing home-specific control variables Xit: Certified staffing ratios it =𝛽𝛽0+𝛽𝛽1ln ( P NC it ) +𝛽𝛽2ln ( P AS it ) +𝛽𝛽3ln ( P S it) +X it 𝛼𝛼+R c𝜓𝜓+𝛾𝛾t+𝛿𝛿c+u it (1) 𝐴𝐴𝐴𝐴 𝑁𝑁𝑁𝑁 𝑖𝑖𝑖𝑖 denotes the case-mix adjusted daily care price, 𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴 𝑖𝑖𝑖𝑖 refers to the daily lodging price paid for accommodation and services, and 𝐴𝐴𝐴𝐴 𝑆𝑆 𝑖𝑖𝑖𝑖 denotes subsidies. The log of daily prices is used to adjust for outliers. Xit controls for nursing-home characteristics like the ownership type and the size of the nursing home. Rc controls for changes in the cantonal price level and income by including the GDP per capita and the cantonal unemployment rate. We also control for changes in cantonal demographics using the share of the population aged 65 and above. ɣt captures year fixed effects, δc refers to canton fixed effects, and uit denotes the error term. While year fixed effects control for changes in the overall economic situation, the canton fixed effects capture potential differences across cantons, such as demographics or the need assessment system. FIGURE 1 Cantonal averages of the yearly variation in price components and registered staffing ratios across nursing homes. The number of registered nurses and care prices are weighted by case-mix. Prices are deflated using consumer price indices (Federal Statistical Office(2021), base year 2012). The size of the circles reflects the underlying number of nursing homes in the respective canton. Cantons: AG=Aargau, AR: Appenzell Ausserrhoden, BE: Bern, BL: Basel-Landschaft, BS: Basel-Stadt, GR: Grisons, JU: Jura, SG: St. Gallen, SO: Solothurn, TG: Thurgau, VD: Vaud, VS: Valais. Source: FOPH(2021), years 2012–2017, own calculations [Colour figure can be viewed at wileyonlinelibrary.com] AG AR BE BL BS GR JU SG SO TG VD VS -4 0 4 8 12 Reg. nurse staffing per 1,000 care days (yearly change, in %) -2 0 2 4 6 8 10 Price for nursing care (yearly change, in %) AG AR BE BL BS GR JU SG SO TG VD VS -4 04 8 12 Reg. nurse staffing per 1,000 care days (yearly change, in %) -2 0 2 4 6 8 10 Price for accommodation and services (yearly change, in %) HEGER Et al. 1625 To interpret the coefficients of the price components as causal effects, the variables must be exogenous with respect to staffing ratios. For the care price, exogeneity follows as we limit our sample to cantons that assign fixed prices per care level. Exogeneity further implies that the change in the regulated price is unrelated to other factors that might drive nursing-home specific staffing levels, conditional on the covariates and fixed effects included. We are not aware about any changes in the pricing regulations, staffing level regulations or definitions of the individuals' care needs between 2012 and 2017, especially not in combination with changes in prices. Our data also show that, within cantons, changes in lodging prices are not correlated with changes in care prices (controlling for covariates and fixed effects, regression results upon request). Subsidies are small on average and are usually paid irregularly to help nursing homes with large investments, for example, in new buildings, or to reduce an overall deficit. We think that potential endogeneity concerns are negligible. Finally, the third price component, the lodging price, may be endogenous due to unobserved demand factors that influence both, lodging prices and staffing ratios positively, such as preferences for high quality and education. Furthermore, since cross subsidizing is officially illegal, some nursing homes do not exploit the resident's willingness to pay higher lodging prices to finance their care costs although they would if it was legal, while others do. Both issues lead to a downward bias of the OLS estimates. Therefore, we first exploit the panel structure of our data and run nursing home specific fixed-effects (FE) regressions. With the FE approach, we exploit within-nursing home variation in lodging prices and get closer to the average treatment effect of the treated (what happened to staffing ratios if there was no cross-subsidization). The fixed effects take out unobserved demand and supply characteristics that vary across nursing homes within the canton, but do not change over time, for example, local willingness to pay, long-term care needs or the nursing home's willingness to subsidize care with private contributions. Additionally, we employ a two-stage least squares estimation approach (2SLS). As an instrumental variable, we use the average lodging price of comparable nursing homes that lie within the same canton, but outside a 15-min driving radius (exclusion radius). The idea stems from the industrial organization literature, where prices are instrumented with common cost-shifters (such as prices of ingredients to the final product or prices from related products in other markets) to eliminate any unobserved demand effects (Hausman,1996; Nevo,2000). We explain the idea in more detail below. Our first-stage estimation is given by ln ( PAS it ) =𝛼𝛼0+𝛼𝛼1ln ( PAS ∗t ) +𝛼𝛼2ln ( PNC it ) +𝛼𝛼3ln ( PS it ) +X it𝛽𝛽+R c𝜓𝜓+𝛾𝛾t+𝛿𝛿c+v it (2) where 𝐴𝐴 𝑃𝑃 𝐴𝐴𝐴𝐴 ∗𝑡𝑡 denotes the average lodging price of the similar nursing homes in the same canton that are assigned to nursing home i and vit denotes the error term. Furthermore, all remaining variables from the second-stage regression are included in regression (2). For identification, we need to assume that 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 ( 𝑃𝑃𝐴𝐴𝐴𝐴 ∗𝑡𝑡,𝐴𝐴 𝑖𝑖𝑡𝑡 )=0 , which means that the instrumented price must not be correlated with the error term in the second stage and must not have a direct effect on nurse staffing other than through the lodging price. The instrumental variable is only relevant if 𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴 ( 𝑃𝑃𝐴𝐴𝐴𝐴 𝑖𝑖𝑖𝑖 , 𝑝𝑝𝑝𝑝𝑖𝑖𝐴𝐴𝑝𝑝𝑖𝑖𝑖𝑖 ) ≠ 0 , that is, α1 must be significantly different from zero. 2.5.1 | Instrumental variable The nursing home market is a very local market and most nursing home residents choose a nursing home close to their original home (Schmitz & Stroka,2020). Nursing homes beyond the 15-min driving radius are unlikely to be direct competitors and do not impact either the price setting or the quality choice of the respective nursing home. The exclusion radius is reduced to 5min within the five cities in our sample since nearly all nursing homes are reached in 15min within cities. By doing this, we exclude nursing homes that compete for the same care-dependent residents. Instead, we use similar nursing homes with respect to (i) size (the number of beds does not differ by more than 10) to capture potential economies of scale, (ii) ownership structure (for-profit vs. not for-profit) to capture intrinsic motivation to provide high quality or to exploit the residents' willingness to pay, and (iii) location (urban vs. rural) to capture location differences in rents. The average lodging price of these similar but non-competing nursing homes serves as our lodging price instrument. Within-canton variation in the lodging price (which is not regulated and can be freely set by the nursing homes) stems from variation in local prices for housing and amenities. By using lodging prices of similar, but not close nursing homes, we eliminate demand shocks (e.g., differences in care needs due to differences in education or family composition) as well as quality preferences that might play a role for the individual's willingness to pay. We rely on the distance as the most important selection criterion that overrules preferences for quality beyond the 15min threshold (Schmitz & Stroka,2020). Calculating the average lodging price of the comparable nursing homes for each nursing home separately and inserting canton and time fixed effects, we exploit within-canton variation in lodging prices across the groups of comparable nursing homes (e.g., bigger nursing homes HEGER Et al. 1632 Lin, H. (2014). Revisiting the relationship between nurse staffing and quality of care in nursing homes: An instrumental variables approach. Journal of Health Economics, 37, 13–24. https://doi.org/10.1016/j.jhealeco.2014.04.007 Mueller, C., Arling, G., Kane, R., Bershadsky, J., Holland, D., & Joy, A. (2006). Nursing nome staffing standards: Their relationship to nurse staffing levels. The Gerontologist, 46(1), 74–80. https://doi.org/10.1093/geront/46.1.74 Nadash, P., Doty, P., & von Schwanenflügel, M. (2017). The German long-term care insurance program: Evolution and recent developments. The Gerontologist, 58(3), 588–597. https://doi.org/10.1093/geront/gnx018 Nevo, A. (2000). Mergers with differentiated products: The case of the ready-to-eat cereal industry. The RAND Journal of Economics, 31(3), 395–421. https://doi.org/10.2307/2600994 OECD. (2021a). OECD.Stat (database). Health expenditure and finance. https://stats.oecd.org/ OECD. (2021b). OECD.Stat (database). Productivity. https://stats.oecd.org/ Philippon, T., & Skreta, V. (2012). Optimal interventions in markets with adverse selection. The American Economic Review, 102(1), 1–28. https:// doi.org/10.1257/aer.102.1.1 Plassmann, H., D'Oherty, J., Shiv, B., & Rangel, A. (2008). Marketing actions can modulate neural representations of experienced pleasantness. Proceedings of the National Academy of Sciences, 105(3), 1050–1054. https://doi.org/10.1073/pnas.0706929105 Reichert, A., & Stroka, M. (2018). Nursing home prices and quality of care Evidence from administrative data. Health Economics, 27(1), 129–140. https://doi.org/10.1002/hec.3518 Rothgang, H. (2020). Entwicklung und Erprobung eines wissenschaftlich fundierten Verfahrens zur einheitlichen Bemessung des Personalbedarfs in Pflegeeinrichtungen nach qualitativen und quantitativen Maßstäben gemäß § 113c SGB XI (PeBeM). https://www.gs-qsa-pflege.de/wp-content/ uploads/2020/09/Abschlussbericht_PeBeM.pdf Schmitz, H., & Stroka, M. (2020). Determinants of nursing home choice: Does reported quality matter? Health Economics, 29(7), 766–777. https:// doi.org/10.1002/hec.4018 Spasova, S., Baeten, R., Coster, S., Ghailani, D., Peña-Casas, R., & Vanhercke, B. (2018). Challenges in long-term care in Europe. A study of national policies. European Social Policy Network (ESPN). European Commission. The Federal Council. (2021a). Federal constitution of the Swiss confederation of 18 April 1999. (status as of 7 March 2021) https://www.admin.ch/ opc/en/classified-compilation/19995395/index.html The Federal Council. (2021b). Federal health insurance act as of 18 March 1994. (status as of 1 April 2021) https://www.fedlex.admin.ch/eli/ cc/1995/1328_1328_1328/de Tong, P.K. (2011). The effects of California minimum nurse staffing laws on nurse labor and patient mortality in skilled nursing facilities. Health Economics, 20(7), 802–816. https://doi.org/10.1002/hec.1638 Tscharnke, K. (2009). Pflegeskandal: Diakonie um Schadensbegrenzung bemüht. https://www.merkur.de/bayern/mm-pflegeskandal-diakonie-schadensbegrenzung-bemueht-304025.html SUPPORTING INFORMATION Additional supporting information can be found online in the Supporting Information section at the end of this article. How to cite this article: Heger, D., Herr, A., & Mensen, A. (2022). Paying for the view? How nursing home prices affect certified staffing ratios. Health Economics, 31(8), 1618–1632. https://doi.org/10.1002/hec.4532 HEGER Et al.