2.1 The DB project connecting the quality and performance of law

The continuing reliance on utilitarian thought aside, there were other major problems in the 2014 DB data revision as well. The 2014 data revision had offered an opportunity to develop the quality TLIs and performance TLIs separately. In fact, there are a number of statements in the DB reports that suggest the performance of law or the quality of law can lag behind each other, and thus can express different empirical facts. For example, with regard to the Resolving Insolvency Indicators, the DB report admits that ‘Many economies have an insolvency system with low efficiency and high quality. These are economies that have well-designed laws but face challenges in implementing them effectively’ and adds that ‘well-designed laws are necessary but not sufficient to achieve efficiency in an insolvency system’ (WBDB, 2014, p. 10).

However, such statements that create a suspicion that the quality and performance of law may not align are quickly dispelled by stronger and more explicit statements that they do. For example, the 2014 DB report also adds to the statements above: ‘On average, though, economies with better-designed laws tend to have higher recovery rates’ and ‘economies that offer a simple, fast and inexpensive process for transferring property are also likely to have a land administration system providing reliable land records’ (WBDB, 2014, p. 10). The 2016 report even goes one step further and contradicts the 2014 report's statement that there are countries with good-quality laws and poor efficiency. It states instead that there are very few economies that are currently in the group of economies ‘where the quality of regulation is high but the processes for implementing it remain complex’ (WBDB, 2016, p. 6). Rather, it insists that ‘those with low regulatory efficiency tend to also have low regulatory quality’ (WBDB, 2016, p. 6).

In fact, the statement that ‘quality and efficiency go hand in hand’ is one of the most recited statements in the DB reports after the 2014 revision (WBDB, 2016, pp. 1, 6–7, 12; 2017, pp. 8, iv). This narrative in the DB reports connects these distinct aspects of the law. It does this in two ways: first, by including certain measurements of performance of law inside the measurements of quality of law and, second, by explicitly and strongly arguing that the quality and the performance of the law are positively correlated – that is, the improvement (or deterioration) in one corresponds to (or, more ambitiously, leads to) the improvement (or deterioration) in the other.

The first way of connection can also be called a ‘measurement trick’. By taking into account some factors that make up a law's efficient performance while measuring their quality, the DB quality TLIs are already to some extent interlinked with the performance of law. This also reinforces the appearance of a correlation between the quality and performance TLIs. For example, the Strength Index – that is, the quality-of-bankruptcy-law measure in the DB dataset – includes the existence or absence of reorganisation bankruptcy procedures, which can increase a country's index score by up to 25 per cent. The Recovery Rate – that is, the performance-of-bankruptcy-law measure in the DB dataset – also rewards having the reorganisation bankruptcy law by up to 30 per cent. Thanks to this trick, the improvement in one indicator is automatically mirrored by the other, which creates an appearance of a positive correlation.

The second way of connecting the quality and performance of law is perhaps more serious, because it is seemingly supported by empirical evidence. Case-studies provide the primary evidence in the DB reports. For example, the 2016 DB report heavily relies on case-studies to show this positive correlation inside land-registry laws and insolvency regimes. The report details that the Danish land-registry system requires very few procedures and time to complete a registry, which encourages people to register their property transfers more accurately – this, according to the report, is evidence that ‘regulatory efficiency and quality go together and in fact reinforce each other in a virtuous cycle’ (WBDB, 2016, p. 6). In the ‘inefficient’ Greek registry system, by contrast, people are discouraged from keeping good registries, and this creates a ‘vicious-cycle’ problem instead. Similarly, the report contrasts the insolvency regimes of Finland and São Tomé and Príncipe, and argues that ‘where there is a good legal framework for insolvency, creditors recover a larger share of their credit at the end of the insolvency process’ (WBDB, 2016, p. 9).

The DB reports also provide evidence based on macro-level data showing a positive correlation between the performance and the quality of law. These are often simple representations of data – what Niemann calls ‘associational charts’ (Niemann, Reference Niemann2013). Niemann argues that these charts present relationship between two variables to offer a possible explanation for an outcome of desire, but they do not have enough statistical power and reliability to show a correlation that can strongly suggest a causation (Niemann, Reference Niemann2013, p. 56). One of the clearest examples of such an associational chart can be found in the 2015 report, reproduced by the same data in Figure 1, which shows a positive regression line formed by Strength Index and Recovery Rate data in 2014 (the first year these data were collected by the WB).Footnote ¹ This simple graph alone led the 2015 and 2016 reports to state that ‘there is a positive correlation between the recovery rate for creditors and the strength of the legal framework for insolvency’ (WBDB, 2014, p. 10; 2016, p. 9). Similar associational charts were also used to suggest that the quality of the land-registry systems positively correlates with the efficiency of land transfers and the quality of court infrastructures with the efficient enforcement of contracts (see other charts on WBDB, 2014, p. 11).

Figure 1.

Reproduced with the DB 2014 data by the author (WBDB, 2014, p. 10).

The rest of this paper will show that, by using associational charts and case-studies, the DB project does not provide real evidence for the existence of a correlation, but rather creates an appearance of correlation between the performance and the quality of law. We cannot dispute each and every case-study or macro-level statistical analysis that the WB uses in these reports; instead, we will focus on the evidence that the WB could collect by measuring the quality and the performance of bankruptcy laws with its Strength Index and Recovery Rate indicators. The seeming connection between the Strength Index and Recovery Rate has been crucial to the WB's argument that the quality and performance go hand in hand.

This paper will also only focus on the DB data from 2014 when running new macro-level tests. This was the first year that the new DB quality-of-law indicators started measuring countries’ scores and these scores were reported in the DB 2015 report. This was a choice to make the presentation of the empirical tests simpler, but also taking multiple years of data did not make any difference to the results. Similar analyses can be run using the following recorded years, or by taking an average of all the recorded years, but the conclusions are the same, since the scores and rankings in these TLIs do not change much over the years. Also, since the 2014 revision, there has not been any important measurement revision in the DB data to warrant updating the results presented in this paper.

3.1 Resolving Insolvency Indicators

The DB's Resolving Insolvency Indicator set is composed of two main indicators: the Recovery Rate and the Strength Index. The Recovery Rate indicator ranges between 0 and 100, and estimates the percentage (or cents on the dollar) of the insolvent firm's assets that secured (financial) creditors can recover through a bankruptcy procedure. The components of the Recovery Rate are the average time, cost and (the likely) outcome of the insolvency procedure, combined with the interest-rate estimate to calculate the depletion of the assets over time.Footnote ² Higher numbers on the Recovery Rate indicate the efficiency gains of the creditors in the process. The Strength Index indicator ranges between 0 and 16, and measures the compliance of the national bankruptcy laws with the WB-recommended best practices on four aspects: the commencement of the bankruptcy process, the management of the debtor's assets during the bankruptcy process, the creditors’ participation during the bankruptcy process and, lastly, the creditors’ rights and duties when there is a reorganisation procedure (WBDB, 2014, p. 96). Compliance in each of these best-practice areas is separately measured: scores in the Commencement of Proceedings Index range between 0 and 3, the management of debtor's assets index ranges between 0 and 6, the creditor participation index ranges between 0 and 4, and the reorganisation proceedings index ranges between 0 and 3. Countries’ Strength Index scores are estimated by a simple addition of these.Footnote ³ Higher numbers on the Strength Index indicate better access of the debtors to the bankruptcy procedure, better protection against the creditors when they can save their businesses and better rights to creditors in controlling and surveying the process (see Table A1 in the Appendix for more details on both indicators and how they are calculated).

The Recovery Rate and Strength Index are connected to each other by both rewarding the adoption of the reorganisation bankruptcy procedure. The base rate of the Recovery Rare is 30 points (or 30 per cent) and the Strength Index is 3 points (or 18.75 per cent) higher for countries with a reorganisation bankruptcy procedure.Footnote ⁴ This is because the Recovery Rate's estimation is based on the assumption that keeping or selling the insolvent companies as ‘going concerns’ (or in operation) is a more efficient outcome, which requires the existence of some sort of a reorganisation bankruptcy procedure (Djankov et al., Reference Djankov2006). This assumption is based on the economic research that shows more recovery for secured creditors in successful reorganisation cases. For example, Bris and others (Reference Bris, Welch and Zhu2006) found that, in the US, the mean Recovery Rate is 69 per cent for reorganisation proceedings, as opposed to 27 per cent for liquidation (also see Menezes, Reference Menezes2014; Armour et al., Reference Armour2015). Similarly, the Strength Index’ Reorganization Proceedings sub-index is calculated as 0 when a country does not have any reorganisation bankruptcy procedure. This reflects the DB's commitment to promoting the reorganisation bankruptcy procedure as a ‘best practice’ (see WBDB, 2014, p. 96; 2016, p. 99), based on the WB's own guide called ‘Principles for Effective Insolvency and Creditor/Debtor Regimes’ (World Bank, 2001). However, despite this common valuation of the reorganisation bankruptcy procedure, this procedure itself is not sufficient to create a direct and strong link between the performance and the quality of the bankruptcy law.

3.2 Testing the macro-level evidence from the WBDB data

Figure 2 shows the number of countries with and without a reorganisation procedure in 2014, broken down into those with an efficient bankruptcy outcome (have the likely outcome of selling or saving the insolvent companies as going concerns) and without an efficient outcome (have the likely outcome of selling them piecemeal through foreclosure or liquidation) and reports their recovery rates.Footnote ⁵ It shows that a large majority (106 out of 167) had some sort of a bankruptcy reorganisation procedure but, in practice, also a large number of these countries (66 out of these 106) could not process their insolvency cases through a reorganisation procedure. This means that, in many countries, companies that should be reorganised by the available reorganisation laws were either falsely liquidated or foreclosed at the end of the bankruptcy process. These countries’ Recovery Rate was also very low (29.7 per cent) and was almost the same as in the countries with lower-quality laws that foreclose or liquidate insolvent companies without a formal alternative (27.18 per cent). In fact, only thirty-three out of 167 countries had the reorganisation bankruptcy best practice and could implement it efficiently at the same time, thus have a high Recovery Rate of 74.46 per cent. This shows that the reorganisation bankruptcy procedure alone cannot establish a strong positive correlation between the Strength Index and the Recovery Rate.

Figure 2.

Insolvency regimes in terms of having reorganisation bankruptcy and getting the efficient outcome of bankruptcy in 2014 (source: WBDB data)

The simple associational charts in the DB reports like Figure 1 do not take into account other factors that can play a role in the performance of law. The statistical analysis presented in Table 2 tests whether including these factors in the analysis changes the appearance of a positive correlation between the Strength Index and Recovery Rate. One of the most important factors omitted by the WB is the quality of the courts system. Scholars have argued that inefficient courts can be significantly detrimental to the bankruptcy process; their slowness (Palumbo et al., Reference Palumbo2013), congestion (Espasa and Esteller-Moré, Reference Espasa and Esteller-Moré2015) and lack of resources and personnel (de Oliveira Gomes et al., Reference de Oliveira Gomes, de Aquino Guimaraes and Akutsu2016) can cause lower rates of recovery and rehabilitation. Considering these, the WB and the IMF also advise countries on reforming and modernising their courts after they reform their bankruptcy laws (e.g. see Rowat and Astigarraga, Reference Rowat and Astigarraga1999; Cirmizi et al., Reference Cirmizi, Klapper and Uttamchandani2012). This factor is accounted for in Table 2 by the Quality of Judicial Proceedings Index, which measures the quality of the court structure and proceedings, case management, court automation and alternative dispute resolution in a court system (WBDB, 2016, p. 31).

Table 2.

Table of regressions

Robust standard errors in parentheses. ***p < 0.01; **p < 0.05; *p < 0.1.

Another important factor that the WB's analysis omits is the legal origin. Different legal traditions – English common law and French civil law – typically have different legal procedures, which are shown to have important effects on legal efficiency (see La Porta et al., Reference La Porta1996; Djankov et al., Reference Djankov2002). Civil-law countries are often thought to have less efficient legal procedures than the common-law countries. To account for this factor, this paper uses the most extensive legal-origins variable compiled and used in the last La Porta et al. work (La Porta et al., Reference La Porta, Lopez-de-Silanes and Shleifer2008).

Another factor that can play a role in the performance of bankruptcy laws is the level of financialisation. The amount of available credit for the companies going through a reorganisation bankruptcy procedure is vital to their survival, and thus the amount of credit that the creditors can recover with this procedure (see e.g. Menezes, Reference Menezes2014). This factor is controlled in Table 2 by the domestic credit provided by finance indicators published by the IMF and WB.

Column 1 in Table 2 shows the positive correlation reported by the WB in Figure 1. This analysis only controls for the size of the economy (using logged GNI variable). Column 2 adds to this analysis the quality-of-courts variable (i.e. Quality of Judicial Proceedings Index), which reduces the effect of the quality of bankruptcy law (Strength Index) over the performance of this law to a statistically insignificant amount. Column 3 further adds the legal origins and the financialisation controls. This model is the most complete model that can be achieved with these variables (as shown by a higher R-squared). It again shows that the effect of the quality of law is insignificant on the performance of law, which is largely determined instead by the size of the economy, the quality of courts, legal origin and amount of financialisation (interestingly, both the British origins and French origins have a positive effect, while the highest positive effect is recorded by the Scandinavian origin). In other words, both Column 2 and Column 3 suggest that the omission of such potentially important factors leads to a misrepresentation of the relationship between the quality and the performance of bankruptcy laws.

Furthermore, the WB's analysis leaves unclear whether the improvements in each sub-index of the Strength Index (measuring and representing different parts of the bankruptcy law subject to substantive quality improvements recommended by the WB) can lead to improvements in the Recovery Rate or only some of these sub-indices are important. Column 4 in Table 2 tests for the importance of these different quality measures separately (instead of their aggregated form, as in the DB reports). This analysis reveals that only one component of the Strength Index – the Commencement of Proceedings Index – might have a positive and significant effect on the performance of law. The WB bases this index on two bankruptcy-law-quality recommendations: both the creditors and debtors should be able to initiate both the liquidation and reorganisation bankruptcy procedures, and the standard test of bankruptcy should be the ‘liquidity test’ – that is, the debtor is generally unable to pay its debts as they mature, rather than the more restrictive ‘balance-sheet test’ (i.e. the liabilities of the debtor exceed its assets). Making these changes should increase the number of bankruptcies filed (both liquidation and reorganisation). Explaining why a numerical increase in filings could increase the performance (efficiency) of bankruptcy laws is a matter for another paper. Yet, this test shows that not all the quality improvements recommended by the WB (in fact, most of them) can have a positive effect on the performance of law.

Lastly, it is important to check whether there are categorical effects over the Recovery Rate – that is, whether countries with different levels of economic development respond to the increase in the quality of the bankruptcy regime differently. Figure 3 shows four separate fitted lines showing the relationship between the Strength Index and the Recovery Rate for the four categories of level of income recognised by the WB (low, low–middle, high–middle and high income).Footnote ⁶ The left-hand side of the chart shows the regression lines without the quality of judiciary control and the right-hand side shows the regression lines with this control. Both sides of the chart indicate that an increase in the Strength Index corresponds to a statistically significantly higher Recovery Rate only for the high-income countries. In fact, for countries with less than a high income (although estimated insignificant in this regression), the correlation might even be negative. This suggests that the positive correlation drawn and presented in the WB DB's association chart for all countries in the dataset might be carrying over the positive correlation for the high-income countries. This means the strongly argued positive correlation between the quality of the law and the performance of the law may not exist, especially for the countries that the WB targets the most as the audience of its DB project.

Figure 3.

Scatter plots of Strength Index and Recovery Rate with separate regression lines for different income groups

Of course, these simple tests on the correlation between the quality and the performance of law are not conclusive and cannot dismiss entirely that such a correlation exists. Reforms in the quality of law and the performance of law can affect each other through other mechanisms. However, these tests are sufficiently sophisticated to show that the statistical evidence provided by the DB project is far from conclusively showing a positive correlation between the quality and performance of law.

3.3 The case of Turkey's WB-led insolvency-law reform

Aside from associational charts, the WBDB reports use country case examples to substantiate their claim that the quality and the performance of law are mutually reinforcing or that they work towards a common goal. In a similar vein therefore, this paper uses a country that experienced a WB-led insolvency-law reform and investigates whether the expected positive effects of these legal reforms have materialised in this country in a way that indicates a positive relationship between the quality and the performance of law. The case of Turkey is illuminating, since it is an important middle-income country that adopted a WB-recommended bankruptcy-law reform in the early 2000s. Turkey's example is different from the example of other countries that the WB uses to substantiate its claim (such as Finland and São Tomé and Príncipe, as mentioned earlier).

Turkey made extensive reforms to its bankruptcy-law system in the immediate aftermath of its deepest financial crisis in 2001, which was triggered by 1999 IMF-led exchange-rate-based disinflation programme (Öniş, Reference Öniş2009). Following the crisis, Turkey signed two stand-by agreements with the IMF (in 2001 and 2005), which led the country to promise a bankruptcy-law reform under the guise of the WB advisers invited to the country. A number of WB advisers supervised and observed the bankruptcy-law reforms, legislated in the Turkish parliament in 2003 and 2004. The single most important component of these reforms was the establishment of ‘Postponement of Bankruptcy’ (Iflas Erteleme) – an in-court reorganisation bankruptcy procedure. This replaced an older and outdated reorganisation procedure that already existed in Turkey: ‘Concordat’ (Konkordato).

Postponement of Bankruptcy has several important aspects that align with the ‘good-practice rules’ identified by the WB guidelines and other IOs. First, the law allows both the insolvent debtors and their creditors to initiate any bankruptcy procedure, regardless of size, sector and debt structure. Second, in line with the WB's recommendations (IMF, 1999, p. 41; World Bank, 2001, p. 29), the law uses a ‘liquidity test’ and a ‘rehabilitation plan’. In the rehabilitation plan, the insolvent company should demonstrate that, although it is insolvent, it is still a viable business and can survive its insolvency by a planned future course of action to rehabilitate its finances and business structure. Third, per these recommendations, the Postponement of Bankruptcy includes ‘interim protective measures’ that prevent a ‘premature break-up’ of assets by the creditors (World Bank, 2001, p. 16) and also protect the estate from the debtors’ transfers (IMF, 1999, p. 23).Footnote ⁷ The court can announce an interim moratorium, which is a period of temporary protection of the debtors’ assets by putting a stay order on all debt-collection proceedings filed against the debtor by its creditors that lasts until the court makes its final decision. During this period, the court also appoints an interim supervising court administrator or a ‘trustee’ to manage the insolvent's assets, whose approval on the management's decisions is necessary for them to come into effect.

However, despite conforming to the WB's recommended best practices, the Postponement of Bankruptcy could not perform very well and was not effective in recovering debt and keeping the insolvents as going concerns. The statistics from the Turkish Ministry of Justice and the data I collected on the cases filed in Istanbul courts in 2015 (see Table 3) show large inefficiencies. First, the number of companies that filed for and were allowed to pursue a reorganisation bankruptcy was very high, which indicated an inability to filter out the unviable businesses in the early stages. Between 2009 and 2016, approximately 583 Postponement cases were filed and 533 cases were resolved each year, which corresponds to seven postponements for ten liquidations. This ratio is much higher than in economies like the US (2.4 in 2016 (Aacer data in 2017)) and the UK (1 in 2015 (The Insolvency Service data in 2017)) and suggests that a typical insolvent company in Turkey is more likely to be allowed to file for reorganisation than its peers in a developed economy.Footnote ⁸ Second, the courts approved the reorganisation plans of only very few companies, which shows that, for the majority of cases that were in reorganisation, the court's resources were spent inefficiently and ineffectively. In the same period, according to the state statistics, the courts nationwide approved only 24 per cent of the cases filed for Postponement.Footnote ⁹ The Postponement filings in Istanbul in 2015 show an even lower rate (see Table 3): among the ninety-five companies in the dataset, only eleven received Postponement and had an open case of reorganisation (11.6 per cent) and the courts decreed liquidation for most rejected companies (about 86 per cent). Furthermore, when these data were being collected (April 2017), the courts still had not made a final decision for forty-five of the cases (47 per cent), which shows how lengthy the trials were.

Table 3.

Case results from 2015 Istanbul Reorganization Bankruptcy Filings

The WB's own calculation of the Recovery Rate in Turkey confirms these findings. According to the WBDB survey, the Recovery Rate in Turkey stood at 18.5 per cent in 2017, which is much lower than the OECD high-income economies (73 per cent). The WBDB survey reports two observations on the Turkish bankruptcy-law system explaining this poor performance, which matches the findings from locally collected data: first, the outcome is almost always ‘piecemeal sale in liquidation after an attempt at reorganization’ and, second, the litigation is on average very lengthy (4.5 years) (World Bank Doing Business 2017 data). The recovery rates are low in Turkey's bankruptcy-law regime because liquidation after a failed reorganisation attempt is less efficient for the creditors than a direct liquidation, since the delays deplete the company's resources and the longer the reorganisation case, the costlier it is for the creditors (Thorburn, Reference Thorburn2000). It is no surprise, then, that the Turkish banking sector has been so strongly critical of the Postponement of Bankruptcy that it demanded substantive changes to the law (The Bank Association of Turkey press conference 2017, read at ANKA, 2007) and that it frequently called Postponement ‘one of the most serious problems for the banking sector’ (Interview with the Participation Banks Association of Turkey Secretary Osman Akyüz, read at T24, 2008).

According to my interviews with the experts in Turkish bankruptcy law,Footnote ¹⁰ there are many potential causes for the poor performance of reorganisation bankruptcy laws in Turkey. For example, the underlying economy is not conducive to reorganisation implementation; the banks and the debtor companies can hardly ever successfully negotiate a substantial reorganisation on the debt, due to both parties’ insufficient funds; banks insist on collateral for debt agreements and prefer confiscating their collateral over saving their loans. Furthermore, the Turkish courts are inefficient, slow and run a large backlog, and judges are generally not trained in detecting and differentiating viable businesses. The interim trustees are similarly untrained in managing businesses, or they are corrupt. While these factors contribute to the reorganisation bankruptcy's inefficiency, they also eliminate the beneficial effects that could be attained from having good-quality bankruptcy procedures.

Even more tragically, the best practices promulgated for the bankruptcy laws further deteriorate the situation. In fact, there are common complaints that the quality improvements in the Turkish bankruptcy law caused the debtors’ misuse. While the Turkish insolvent debtors gained greater access to bankruptcy procedures with the reform, they often used this right to empty the companies’ remaining assets while holding off their creditors, in the absence of a sufficient court overview of the process. They also used the liquidity tests to show solvent companies insolvent, which serves towards pushing those companies’ creditors into renegotiating the debt in more favourable terms to the debtors. The rehabilitation plans can similarly be used to paint a more optimistic assessment of the future prospects of the company by exaggerating its past achievements and the viability of its future plans. Lastly, interim protective measures, and especially the interim moratorium, were detrimental to the success of the Postponement of Bankruptcy. Almost every company that applied for the reorganisation bankruptcy law received an interim moratorium at the beginning of the process in order to protect it from its creditors. A company could easily get three to four years of interim moratorium while awaiting the decision of the court and use this period to pressure their creditors.

As a result of this large-scale misuse and inefficiencies, the Turkish reorganisation bankruptcy law was annulled in 2018. Turkey today uses its more outdated reorganisation bankruptcy procedure, ‘Concordat’, which is stricter in terms of the rights of the debtors and obligations of creditors. For example, it demands that a large number of creditors consent to the reorganisation of the debtor's assets before any protection can be issued (Eroğlu, Reference Eroğlu2005). These restrictions are contrary to the current recommendations of the WB to improve the quality of bankruptcy laws. While there are also efficiency problems with this current law (especially in the current economic crisis that Turkey is going through), the country's reversal of its WB-led reform in the quality of bankruptcy laws suggests that such reforms are not always beneficial to the implementing developing economies.