Low risk, high reward? Repeated competitive biddings with multiple winners in health care

Public authorities in the European Union spend around 14% of the GDP on the purchase of services, works and supplies using public procurement [1]. There is an increasing trend also among health and social care organizers to use competitive elements such as competitive biddings [2]. Well-known imperfect market conditions such as uncertainty and asymmetric information characterize health care markets [3]. Therefore, several issues need to be considered when competitive biddings are used in health care. For example, it is typical to organize the competitive biddings in a repeated manner and select multiple service providers for each contract period. Meanwhile, many patients may receive the service continuously and prefer to visit their usual provider. Competitive biddings give information about the right price-level, but the procurement may also change service providers and end long-lasting relationships between patients and providers. Even though these properties are well-known, only very few studies have analyzed the properties of competitive biddings in health care.

In this study, we analyze competitive biddings that were organized by the insurance districts of the Social Insurance Institution of Finland (Kela) in 2003, 2006, 2010, and 2014 to acquire providers for physiotherapy service for disabled individuals. The districts were responsible for acquiring multiple providers based on their quality–price ratio and the local demand for the service. However, the districts did not have strict budget constraints and historically only very few districts rejected any providers. This setting provides an interesting opportunity to examine the properties of repeated competitive biddings with multiple winners in health services. In the main analysis, we study the effects of the risk of rejection on prices. We also perform counterfactual policy analysis, to examine the direct fiscal effects of implementing a systematic capacity-based rule of provider acceptance.

First, we present descriptive evidence and show that the overall price-level of the bids for a 45-min physiotherapy service increased from an average of 33 euros in 2003 to 58 euros in 2014. Meanwhile, the highest bid increased from 55 to 116 euros. Then, we examine more closely the effects that the risk of rejection had on prices in the 2010 and 2014 competitive biddings. Our main data sources are quality–price scoring tables collected from the insurance districts, merged with patient-level register data on patients’ choice of provider in 2006–2015. We measure the risk of rejection using geographical variation in the insurance districts’ rejection rates in previous competitive biddings. We also analyze whether providers’ previous distance to the rejection threshold had an effect on their prices. In our analysis, we control for market competition, using measures that are based on the value that each provider brings into the insurance district’s network, as well as several other provider and area-level factors.

The results show that rejecting at least one provider decreased prices by more than 5% in the next competitive bidding. The effect is stronger in 2014 than 2010, which suggests that the providers learned the institutional practices of the procurement procedure. We also find that providers which were located further from the rejection threshold increased their prices more heavily in the next competitive bidding. This indicates that providers learned to calculate their optimal bids, which further increased the overall price-level. These results suggest that implementing a systematic capacity-rule of acceptance in each district would have lowered the overall price-level of the service. We simulate the effects of counterfactual scenarios where providers are accepted based on their annual capacity and the local demand for the service. The analysis shows that the regulation would have decreased the costs especially in the 2014 procurement. However, many patients would have been forced to switch their usual provider, which is problematic in a service that is based on continuous relationships between patients and providers.

The study is related to several distinct strands of literature. First and foremost, the study is related to the use of competitive biddings in health care. We are the first to provide empirical analysis regarding competitive biddings in a health service, which patients receive continuously and for which multiple service providers are acquired in repeatedly organized procurements. Theoretical literature suggests that competitive bidding can be a powerful mechanism to decrease the health care expenditure, but their design and implementation must be done carefully [4, 5]. Empirical literature on competitive biddings in health services is very scarce and comes mainly from the analysis of competitive biddings in the US Studies have shown that insurers can use their market power for higher bids in the competitive biddings of Medicare [6, 7]. Similarly, the results of this study show that price bids have been higher in less competitive areas.

The study also contributes to the literature on the effects of competition on prices in health care. For this purpose, the Finnish physiotherapy markets offer an excellent setting because the market consists of a large number of small private providers. Most of the previous literature has documented that competition decreases prices in hospital services [8]. Pekola et al. [9] analyzed the effects of competition on quality and prices using the same setting with a sample of providers in the 2010 competitive bidding. They found that competition had a weak negative effect on quality but no effect on prices. This study extends their work by taking more precisely into account the historical and institutional setting as well as using a much richer provider and patient-level data.

The study also relates to provider contracting in health care and the side-effects of narrowing the network from which patients can choose their provider. Higuera et al. [10] show that narrowing the network can reduce costs, but patients are willing to pay for a wider network that includes their usual provider. Similarly, our study illustrates the simple trade-off between economically efficient procurement and continuity of care at the patients’ usual provider. Finally, the study provides evidence regarding patient choices and provider quality. Our results are similar to previous literature [11‐12], showing that physiotherapy patients choose large, high-quality providers within short distances.

Many countries have implemented policies that increase competition among health care providers [13]. The main purpose of the reforms has been to improve the efficiency and quality of the services. Three main types of provider competition have been presented in the health economics literature: competition in a market, competition for a market and yardstick competition [2]. This paper considers a case where competition both in and for a market is present. Competition in a market usually means that providers compete with quality attributes to attract patients, and the money follows the patients to their selected provider. In this case, prices are often fixed, and the organizer needs to determine the appropriate price-level. To choose, patients should have several alternative providers and access to quality information. Nevertheless, patient choice has become very common in primary health services in European countries [14], and there is increasing empirical evidence showing that quality influences patient choices [11, 15].

Competition for a market means that several potential providers compete for the right to provide services or goods. The purchaser selects one or more providers in, for example, a geographical area. Common examples include the purchasing of pharmaceutical products or hospitals competing to be included in an insurer’s network of providers. Competition for a market requires that the service organizer can describe the services or goods in an accurate and verifiable way [2]. Providers and their prices can be determined, for example, using competitive bidding or bargaining between the purchaser and provider. Compared to fixed prices, competitive bidding provides information about both the prices and the providers that deliver the services at the lowest prices [16]. Therefore, competitive biddings have gained an increasingly important role in discussions about the future financing of health care services [7, 17].

Public procurements and auctions often work very well, but their design must be sensitive to the details of the institutional setting [18]. According to McCombs and Christianson [19], competitive bidding where multiple providers are selected has following four main advantages: first, selecting more than one provider gives flexibility and ensures service availability in cases where a provider exits the market or the demand suddenly increases during the contract period. Second, it might encourage more providers to participate in the procurement, because it increases the probability of being included in the pool of providers. Third, it provides patients a larger degree of choice from the pool of providers, which in turn might increase quality competition in the market among providers. Fourth, accepting multiple providers might ensure market competitiveness in subsequent competitive biddings by preserving viable competitors.

McCombs and Christianson [19] also point out three potential disadvantages: first, in areas where the number of potential providers is very small, there are only weak incentives to submit low bid prices, because the probability of losing is very low. Second, adverse selection problems under a per-episode reimbursement scheme may occur, because the price does not vary but some patients may require more intensive care. Therefore, some providers might try to avoid taking the most demanding patients. Third, accepting a large number of providers might decrease providers’ expectations of number of potential patients and gains from economies of scale, reducing the number of low price bids.

McCombs and Christianson [19] also discuss how the line between accepted and rejected providers should be drawn. A potential procedure is to use a capacity-based rule of acceptance, which requires that all providers submit a maximum capacity of service they can provide during the course of their contract. Accepted providers are then selected in an ascending order of bids, based on the possible quality–price rule, until the target capacity is reached. Under these conditions, the procurement results in a number of providers that ensures access to the services and the long-run market competitiveness. However, if a strict and clear rule of acceptance is not used, providers have strong incentives to “game” the system by submitting bids that are higher than their costs but lower than their estimate of the lowest possible rejected bid.¹ McCombs and Christianson [19] hypothesize that the “gaming” will eventually increase the overall price-level of the bids. Our study empirically tests this hypothesis.

The institutional setting of the study is an individual outpatient physiotherapy service, which is part of the intensive medical rehabilitation services financed and organized by Kela. The physiotherapy is intended for disabled persons under 65 years of age, who face problems managing daily activities and fulfill the criteria defined by law. The basis of the service is a written rehabilitation plan that is drawn up with a physician for 1–3 years at a time. Patients do not pay any out-of-pocket payments and have a free choice of provider from the pool of accepted providers. Typically, patients receive sessions that last for 45 or 60 min, a couple times a week for several years. The physiotherapy is received either at the provider’s facilities or at patient’s home. 14,756 patients received the service in 2015, and the total costs were 73.5 million euros [21].

Kela acquires the service from private physiotherapy providers. Physiotherapy markets are among the most competed health services in Finland. Based on Statistics Finland’s registers there were 2632 independent physiotherapy providers that had 3655 employees and annual turnover of 302 million euros in 2015. Meanwhile, Kela purchased the service from 1253 different providers in 2015. Thus, around 48% of all physiotherapy providers in Finland are Kela’s service providers and the service covers around 24% of the sector’s annual turnover. Overall, Kela is the largest single financer of different rehabilitation services in Finland. Other large purchasers of physiotherapy services for different patient groups are municipalities, the occupational health care sector and insurance companies. Compared to the physiotherapy services organized by municipalities and other financers, patients who receive the service organized by Kela have more severe disabilities and require more intensive therapy.² Individuals can also visit private physiotherapists by paying out-of-pocket payments.³ Usually patients do not simultaneously receive the rehabilitation services that are financed by different organizations.

Kela’s insurance districts are responsible for organizing the service for the local population. The districts acquire the service from private physiotherapy providers using a competitive bidding,⁴ which they have organized in a similar and predefined manner since 2003.⁵ Most of the districts negotiated prices directly with the providers prior to 2003.⁶ The three latest competitive biddings were organized in 2006, 2010 and 2014 for the contract periods 2007–2010, 2011–2014, and 2015–2018. Figure 1 presents the annual nominal costs of the service and the number of patients in 2002–2018. The figure shows that while the number of patients remained relatively stable before the latest contract period, the annual nominal costs more than doubled. The largest cost increase took place in the first year of each contract period, when new prices were set. The number of patients has increased by around 10% during the latest contract period. However, there has been no increase in total costs. This indicates that the annual number of physiotherapy sessions per patient has decreased, most likely as a result of financial pressures caused by the rising costs.

The procurement process begins with a request for tenders. Providers provide information on their quality, annual capacity and set a price for a 45-min physiotherapy session in their tenders. The tenders are submitted to the insurance district where the providers are based. The districts evaluate the tenders and rank the providers that meet the minimum criteria based on their quality–price scores.⁷ Each district decides a rejection threshold based on the capacity of the providers and estimated local demand for the service. The district manager approves this threshold. Providers above the threshold are offered a 4-year contract, and providers that sign the contract form a pool of providers from which patients can freely choose their provider. However, quality information on the providers is not publicly available. Providers are paid for patients’ visits based on their accepted prices. Since the 2006 competitive bidding, the districts have sent the quality–price ranking lists to each provider that submitted a tender in that district.⁸ Providers were also always given information about the quality score rules, and they were able to calculate their points from the information they provided to the district. Figure 3 (in the Appendix) presents an example of the quality–price score table that was used in the Espoo district in 2014.

Even though the districts were expected to accept providers based on local demand and providers’ capacity, the most common rejection threshold has been below the provider ranked last. Figure 2 shows the districts that rejected at least one provider in 2006, 2010 and 2014. In 2006 there were five districts that rejected 11 providers, in 2010 only three districts that rejected seven providers and in 2014 six districts that rejected 28 providers. Altogether only 1.2% of the providers got rejected in the three studied competitive biddings (46 out of 3769). Most of the rejected providers were single outliers, which means that they either offered a significantly high price or had a very low quality. For most of the providers the risk of being rejected was very low, which enabled them to raise prices after realizing this institutional practice. All districts acquired multiple times their required capacity in every competitive bidding. For example, in 2014 the total capacity of the accepted providers was 50,917 whereas only 14,671 patients received the service.

Based on interviews with Kela procurement personnel, there were four main reasons why the districts did not reject more providers: first and foremost, Kela is obliged by law to organize the service nationwide for all eligible patients, and the districts have needed to ensure the availability of the service also in rural parts of the country. Second, the district managers did not have a budget constraint or price limit when they decided on the threshold. Third, Kela has emphasized patients’ freedom to choose from a large pool of providers. Finally, the districts have wanted to avoid a situation where a large number of patients are forced to switch providers because their previous provider was not accepted in the new pool of providers. In conclusion, the district managers did not have proper incentives to reject providers because no financial pressure existed, and rejections would have likely resulted in an increase in administrative work and negative feedback.

We analyze the effects of implementing a capacity-rule for acceptance in the 2014 competitive bidding for the 2015–2018 contract period. We investigate two capacity-rules that are based on the number of patients in the district added by 10 or 100 percent.¹⁷ From a practical point of view, the scenarios take into account potential market exits and increasing demand during the contract periods. We perform the analysis using the following three steps: first, we create the counterfactual pools by including all providers in the quality–price lists until the capacity rule is reached.¹⁸ Second, we examine how patients distribute among providers in the counterfactual pools. We predict the choice of provider among patients who are required to switch, because their initial provider is not included in the counterfactual pool. For the predicted choices we use the estimates from the choice model in 2015 (see Table 4). In these predictions we replace the variable capacity with providers’ remaining free capacity, to take into account that switching patients would be less likely to choose a provider with only little free capacity. The provider with the greatest probability is considered as the chosen one. In the final step we calculate expected annual costs and travel distances per patient at their chosen provider.

Table 7 presents the results of the counterfactual scenarios calculated in 2015. Implementing a 100% capacity-rule results in 586 accepted providers and 601 rejected providers. The average price of the accepted providers is 2.3 euros lower and the average quality 3.6 points higher compared to the actual pools. Because a large number of providers are rejected, 35% of the patients are required to switch providers. However, compared to the actual choices in 2015, an average patient visits a provider with 2.4 points higher quality located only 0.7 km further from their home. Patient-level annual costs are on average 104 euros less than actual costs in 2015, and the total annual costs decrease by 1.35 million euros. This represents annual savings of 2% points. Table 7 also shows estimations for the stricter 10% capacity rule, where only 353 providers are accepted and 60% of the patients are required to switch providers. An average patient visits a provider with 3.7 points higher quality located 3.7 km further away from home, compared to actual choices in 2015. Implementing the rule results in annual savings of 2.3 million euros.

Our results show that implementing a capacity-rule would have resulted in large fiscal savings and higher quality of care in the 2015–2018 contract period.¹⁹ On the other hand, many patients would have been required to switch providers and travel longer distances. Our analysis for the direct fiscal savings can be viewed as lower bound estimates, because rejecting a large number of providers would have probably kept the initial level of the bids lower. Also, the two districts that piloted a fixed price were also excluded from this analysis. Further, the districts have a legal possibility to negotiate direct contracts between patients and providers located in areas with few alternatives. Using these direct contracts can further reduce unnecessary forced switching and increased travel distances in cases where a provider is rejected from the pool of providers but might be necessary for a small portion of patients nearby.

Competitive bidding can be a powerful cost containment mechanism in health services. Even though the use of competitive biddings has increased, there is very little empirical evidence regarding the characteristics of regular competitive biddings in health care. In this study we analyze competitive biddings that were organized repeatedly every 4 years by Kela’s insurance districts to acquire multiple providers for a physiotherapy service in their area. The districts rejected only very few providers because of a lack of financial incentives to use a systematic acceptance rule, and because the districts did not want to terminate established relationships between patients and their usual providers. We analyze whether these features had an effect on providers’ bidding behavior.

Our descriptive analysis regarding the competitive biddings in 2003, 2006, 2010 and 2014 shows that the overall price-level and dispersion of the bids increased heavily during the period. The regression analysis provides further evidence regarding bidding behavior. We show that providers that are located in areas where at least one bid was rejected in the previous competitive bidding offered 5.5% lower prices in the 2014 competitive bidding. Further, we show that providers that were far above the rejection threshold increased their price bids more than providers closer to the threshold. These results indicate that providers learned the institutional features of the competitive bidding and started to behave more strategically. Finally, we perform counterfactual analysis, which shows that using a systematic 100% capacity-rule of acceptance would have resulted in direct fiscal savings of at least 5.4 million euros in the 2015–2018 contract period. Even though rejecting a larger number of providers might have resulted into forced switching of providers and longer travel distances, patients would have also received a higher quality of care.

Our study offers some important policy implications in regularly arranged competitive biddings with multiple winners in health care. We show that unless a clear rejection threshold is used, the competitive pressure regarding prices does not exist. Thus, the overall price-level increases when providers eventually learn that the risk of rejection is low. Applying a systematic capacity-rule of acceptance has two main features that increase efficiency of the services. First, the capacity-rule brings positive dynamic effects because the competitive pressure increases. Even though our analysis focuses only on price competition, it is likely that also quality competition increases as a result of a higher quality–price threshold for acceptance. Second, the capacity-rule also has direct effects because only the most efficient providers with the greatest quality–price ratio are accepted in the pool of providers.

The results raise some challenges for practical decision-making. It might be difficult for the service organizer to suddenly reduce the number of accepted providers for two main reasons. First, rejecting providers might end a large number of existing relationships between patients and providers. Second, it is likely that the travel distances would increase as a result of the rejections. In 2018 Kela renewed its procurement practices. Five large districts that organized the procurement were instructed to accept providers based on the number of patients in their area and the capacity of the providers added by around 10%. Thus, the main policy recommendation of this paper was placed into practise. However, Kela did receive a lot of negative feedback. One possible solution to avoid the negative features is to apply a fixed-price and accept all providers that meet the minimum quality standards. This requires information about the supply-side costs to determine the right price-level. A fixed price might also not encourage providers to invest in their quality beyond the minimum necessary level, unless patients have freedom of choice and quality information about the providers.