Assessing the relation between financial performance and long-term bank loan interest rates for healthcare providers in the Netherlands: a panel data analysis

We assessed whether providers with good financial performance obtain relative lower interest rates on their loans. This would stimulate providers to improve financial performance. Analysis consisted of two steps: (1) trends in interest rates per sector, total capital investments per sector and financial performance per sector; (2) conducting a pooled regression analysis to investigate the relation between the financial performance and interest rates for individual healthcare providers.

The data used in this study were collected from the DigiMV database, a publicly available collection of annual financial reports of Dutch healthcare providers [24]. Publishing annual reports is mandatory for each provider under the Healthcare Admission Act (HAA) (Wet Toelating Zorginstellingen). The healthcare providers are obliged to upload their data in unified formats and in certified annual reports. Small providers (under 10 employees) are exempted. This data covers over 95% of providers in medical and long-term care. Included sectors are general hospitals, university medical centers, nursing homes and home care, mental healthcare, disability care, independent treatment centers and revalidation care. Continuous variable outliers of financial ratios selected through analysis (1.5 interquartile range below first and above the third quartile) were inspected on correctness and transformed where applicable. Transformation occurred based on the variable value presented in the annual financial report. String-based variables were unified by a unification key and manually checked to correct for spelling (e.g., abbreviations and capital use). Data were analyzed using RStudio (version 4.2.0) [25].

We plotted median provider interest rates over time per sector and compare these to the Dutch 10-year government bond interest rate. This demonstrated the opportunity costs of loans. Loans provided to the Dutch government are perceived as relatively risk free, represented by a triple A rating [26]. We defined the interest charges added to this risk-free interest rate as the risk surplus, which is related to the perceived risk of the capital investment by the lender, i.e., the risk of lender bankruptcy relative to the Dutch government [27]. Additionally, we separated loans provided by the HGF, as this was expected to result in lower interest rates.

Second, we plotted total capital investments and financial performance of healthcare providers who obtained capital loans over time. Total capital investments include short-term investments in capital goods, such as computers and diagnostic devices, as well as long-term investments in buildings and facilities. Financial performance was plotted for each indicator separately as well as for a composite score of all indicators. To construct a composite score, we transformed financial ratios to a standard-normal distribution (Z-score transformation) and calculated the mean z-score per provider, reflecting a financial benchmark score (positive if above average, negative if below average).

We utilized a subset of the ratios commonly used by Dutch banks to measure the financial provider performance. This set of ratios was based on a factor analysis performed by Zeller et al. [28], amended by Dutch literature (See Table 1) [18, 29]. To assess profitability characteristics, we used the ratios ROA, TMAR, OMAR and EBITDA. We used the solvency ratio ‘equity ratio’ (ER) as an indicator for financial reserves in the organization. This is probably the most important criterion for providers to get access to the underwriting by both the Dutch HGF and the commercial banks (minimal ER of 20–25%) [18, 29]. Liquidity is another characteristic which HGF clarifies as important. We included the current ratio (CR) and the ‘days cash on hand’ (DCH). Finally, to assess fixed assets efficiency, we utilized both the fixed assets turnover ratio (FATO) and the total assets turnover ratio (TATO). These ratios show how efficiently the organization can generate revenue from their assets. Due to lack of data availability, we excluded the characteristics ‘fixed asset age’ and ‘debt coverage’ from our framework. Moreover, ‘debt coverage’ brought inherent statistical problems, as our aim was to investigate the correlation with the interest rates. We adjusted for loan characteristics affecting the interest rate beyond the financial performance of the healthcare provider, such as the principal amount, the loan period and the method of repayment, as shown in Fig. 1.

Figure 1 demonstrates the conceptual approach of this study. In reality, other (qualitative) factors may impact the bank policy regarding interest rates as well, such as assessment of the investment plans and competition with other nearby providers.

Reduced-form model:

Expanded model:

We mapped the changes in interest rates for the entire healthcare sector (including university medical centers, general hospitals, disability care, mental healthcare, revalidation care and independent treatment centers) in Fig. 2. We clearly demonstrated a declining trend in interest rates, from 4.5% in 2007 to 2.5% in 2019. However, the interest rates on 10-year government bonds showed even larger declines of 4.5% in 2007 to 0.0% in 2019. The interest rate spread between long-term loans and government bonds has thus increased since 2007. This could signal for example that commercial banks rate healthcare loans as increasingly riskier, or that banks receive additional markups on healthcare loans unrelated to the level of risk, such as growing market power of lack of competition.

Investments have declined over the period 2007–2019 (Fig. 3). In 2007, approximately 10% of total revenue was used for investments, compared to only 5% in 2019. About half of the investments in 2007 were financed through capital loans, and this has increased slightly from 2008 onward. This may indicate that retained earnings were used less for investments, and more so for improving the financial position (ER). Net investments were indicated by mutations in fixed assets, which express the value of total investments after depreciation.

Figure 4 shows the financial Z-composite score. The Z-composite score shows the relative financial performance of an organization compared to the average of the entire sector. ROA, EBITDA margin, ER, CR, DCH, FATO and TATO are included in this score. Financial positions increased across the board over the period 2007–2019, although significant variation remains (Fig. 4). A separate trend analysis of the ER iss included in Fig. 5, which demonstrates that financial reserves have doubled over the study period. This is in line with trends in investments in the healthcare sector (Fig. 3), as investments are increasingly financed through capital loans, rather than financial reserves.

While overall trends showed a discrepancy between both higher surpluses on interest rates on the one hand and an increased financial performance on the other hand, this could be due to composite effects (e.g., ecological fallacies) or confounders. We tested whether an association exists at the provider level, i.e., whether good financial performance pays off in terms of lower relative interest rates. This was tested by pooled linear regression, as shown in Table 2. These results demonstrated that financial performance of healthcare providers had no significant relation with the interest rates on long-term banking loans, neither in the reduced-form model or in the expanded model. Robustness checks found either no correlation or significant correlation with wrong signs for individual indicators (Supplementary material 1). For example, a higher EBITDA was associated with higher interest rates.

The results for the expanded model show that the duration of the loan had a positive association with interest rates, suggesting that an increase in loan period duration can result in higher interest rates. The size of the healthcare organization in terms of revenue is associated with lower interest rates. This is confirmed by the relatively low interest rates that university medical centers paid, compared to general hospitals and other providers. Moreover, the results indicated that a HGF guarantee yielded a significant discount on the interest rate over the studied time period. Finally, we observed significant differences between individual banking organizations. A foreign bank had significantly lower interest rates as compared to national banking organizations.

This study is the first that aimed to relate commercial healthcare interest rates to provider financial position over longer time periods, rendering evidence on private capital market performance to fund a fully non-profit healthcare provider sector. Our research provided strong evidence that the capital market reform did not live up to its expectations in improving financial performance of healthcare providers. However, a few limitations should be noted. First, our selection of financial ratios was based on the analysis of Zeller et al., supplemented with commonly used ratios in the Dutch healthcare sector. Although extensive, these ratios might not fully capture the financial performance of healthcare providers. Second, we only observed loans that were realized, introducing selection bias toward successful bargaining outcomes. If providers with low financial stability are unable to obtain loans, market discipline would be present but would not be detected in our current research design. Third, banks might outperform the government in evaluation of healthcare business plans, irrespective of financial position. This would induce efficiency in capital use in healthcare which is not included in our design. However, the decline in healthcare investments can be attributed to a more stringent evaluation process of capital loans by banks. Fourth, competition among providers was not taken into account, which may influence interest rates. For example, hospitals with few competitors nearby may have a higher bargaining position irrespective of financial position. Finally, the publicly available dataset used in this study is self-reported by healthcare organizations. Corrections were applied to mitigate input errors, but some small errors may remain. Given the large set of observations, however, the risk of distorting the analysis is relatively low.

In the Netherlands, healthcare capital is predominantly financed through bank loans. This is potentially an interesting financing model for countries that predominantly rely on government financing. The Dutch case reveals that, under certain conditions, a bank finance reform may reduce investments in healthcare, and at the same time may increase the cost of capital. While loan warrants, such as those provided by the HGF, may reduce the cost of capital, they may be insufficient to reward providers for financial scrutiny. Increased competition in the banking sector may reduce the cost of capital while increasing the incentives for providers to increase their financial position. Governments may also allow for alternative sources of capital (e.g., equity capital) to increase competition. Some studies hint that private equity may increase total health expenses [42], although evidence is mixed [43].

The predominant position of three Dutch banks in healthcare financing begs the question to what extent these banks affect healthcare policy and planning. For example, banks may have a major role in hospital bankruptcy decisions, mergers, or primary care substitution policies. Additional research is needed to assess the role of banks as stakeholder in policy formation.