Recently, various commentators have argued that high non-performing loan (NPL) stocks can impair the monetary policy transmission mechanism by limiting banks’ lending ability.1 This column casts doubts on this thesis, showing that it lacks a solid foundation – on either theoretical or empirical grounds.
NPLs are relatively opaque and heterogeneous, and therefore difficult to value. Also, they typically do not yield a steady return; thus, other things being equal, banks with large NPL holdings are less profitable, and pay a risk premium on capital and liquidity markets. Indeed, the Single Supervisory Mechanism has fostered a reduction of the stock of NPLs in banks’ balance sheets, with very good results. I argue that although there are various reasons to support this policy, not all of them are well-grounded.
There is no theory suggesting that high NPL stocks impair credit allocation
To my knowledge, there is no clear theory suggesting that high volumes of NPLs impair the credit allocation mechanism. The Bank Lending Survey (an EU-wide survey on credit conditions among the main European banks) offers useful insight into this issue.2 Banks are asked inter alia to rank the following drivers of their decision to tighten/ease credit supply – cost of funds and balance sheet constraints; risk perception of borrowers; pressure from competition; and risk tolerance. Interestingly, this list does not include NPLs. Indeed, the BLS survey questionnaire points to an indirect effect – NPLs are opaque and difficult to value; also, their yield is low, or zero. Hence a bank saddled with high NPL levels will be perceived as relatively risky, and may experience difficulties accessing liquidity and capital markets. Other things being equal, these weaknesses will be reflected in a bank’s lending supply.
The argument is correct, but it works only if the bank is perceived as weak. These channels may be dampened, or neutralised altogether, if the bank is sufficiently profitable and/or capitalised. Furthermore, this argument is not specific to NPLs; it applies in general to assets with similar features (high opacity and valuation uncertainty, low liquidity and yield, such as certain Level 2 and Level 3 assets) (Potente et al. 2018), and to banks with some other publicly-known important weakness (e.g. governance). Third, weak balance sheets could in principle induce banks to lend more, rather than less, following a ‘gamble for resurrection’ type of logic. Indeed, risk shifting has been identified as a key incentive for banks to invest in risky bonds during the European sovereign debt crisis (e.g. Acharya and Steffen 2015, Altavilla et al. 2017, Farhi and Tirole 2018); high NPLs could create a similar incentive for banks to increase their credit supply.
Much of the support for the view that a high stock of NPLs can impair the credit allocation mechanism originates from Japan's lost decade. The idea is that banks trying to keep ‘zombie firms’ afloat will be prevented from lending to healthy firms, misallocating credit and generating negative consequences on economic growth (e.g. Caballero et al. 2008). However, the focus of this literature is on ailing firms, and not on NPLs per se.
NPLs include loans to borrowers that are still a going concern – and can, if properly managed, go back to a performing status – as well as loans to firms that are no longer active. Most transactions so far have involved loans of the latter type. In this market, buyers actually acquire a stake in an insolvency procedure, and make money mainly by acquiring, restructuring, and selling the collateral. Thus, these NPL sales do not free banks from zombie firms.
Selling NPLs pertaining to ailing firms would eliminate the zombie lending problem. However, since sources of external finance alternative to banks are still underdeveloped in the EU, these sales could also trigger the bankruptcy of firms with a potential to recover. Evidence available at the Bank of Italy suggests that a non-negligible share of loans to ailing firms go back to performing status after a few years. Good NPL policies should be about maximising the ‘cure rate’, not about getting rid of NPLs at all costs. In sum, the literature on zombie lending can hold lessons for the NPL problem, but addresses a distinct phenomenon.
A version of the argument used by advocates of fast reduction of the NPL stock is that NPLs consume capital, reducing banks’ ability to lend (e.g. Aiyar et al. 2016). This argument overlooks the fact that NPL sales – de facto the only way to effect a rapid reduction of the stock – typically takes place at prices that are much lower than the book value.3 This generates a loss that more than offsets the positive effect on capital ratios generated by the reduction of RWAs.
In sum, the link between NPLs and credit dynamics is an indirect one, working through the negative effects of NPLs on profitability, and on the cost of liquidity and capital. While these effects are more than sufficient to warrant the current strong policy focus on NPL reduction, a serious theoretical analysis of the relationship between NPLs stocks and credit dynamics seems missing. The empirical evidence, to which I turn next, is relatively scarce and does not provide clear-cut conclusions.
The empirical evidence on the issue is scant and inconclusive
The stock of NPLs of Italian banks began to increase with the global recession, in 2008–09, and peaked in 2015 (Figure 1). The figure shows a coarse negative correlation between the stock of NPLs and the growth rate of credit to the private sector over the 2008–2016 period. However, such correlation does not warrant conclusions about causality. Identification is a key issue – NPLs rise in countries and periods where economic activity, and hence credit, stagnates or contracts.
Figure 2 relates an indicator of difficulty to access credit to the flow of new NPLs. Strong co-movements between the two series are readily apparent (their correlation is 0.83). The credit indicator is lagged one year, showing good leading properties for new NPLs. An interpretation of this evidence could be the following—seeing a deterioration (improvement) in the economic outlook, banks begin to tighten (loosen) credit conditions; about a year later, as the economy deteriorate/improves, the flow of NPLs follows suit. Comparison of Figures 1 and 2 clearly shows that in the years around 2015, when the stock of NPLs peaked, the difficulty of accessing bank credit was reaching historical minima. This evidence is not in line with the view that a high stock of NPLs can impair banks’ capacity to issue credit to the economy.
Figure 1 NPLs of Italian banks and bank lending to the private sector
(share of NPLs over outstanding loans; growth rate of bank lending; percentage points)
Source: Bank of Italy.
Notes: Data for 2017 are provisional. Bank lending to the private sector includes households and non financial firms.
Figure 2 NPL flows and difficulty of accessing bank credit by non-financial firms
(percentage points; diffusion index)
Source: ISTAT for the indicator of difficulty of accessing bank credit. Bank of Italy for the new NPL rate.
Notes: The new NPL rate is computed as the ratio between the value of new loans entering the NPL status in a given quarter and the stock of loans at the beginning of the quarter, annualised. The Difficulty of accessing credit is a diffusion index derived from a survey conducted by the Italian statistical Office (ISTAT) on a sample of non-financial firms. See Angelini (2018) for details.
Accornero et al. (2017) study the influence of NPLs on the supply of bank credit to non-financial firms. They review several papers on the issue and argue that identification problems are pervasive. To address these problems they employ an extensive dataset on borrower-level loans in Italy between 2008 and 2015, using time-varying firm fixed effects to control for shifts in demand and changes in borrower characteristics. They find that, although the exogenous emergence of new NPLs and higher provisions can reduce the supply of credit, the NPL ratios per se have no impact on the banks’ lending behaviour. In particular, firms with active relations with more than one bank were treated in the same way by their lenders irrespective of their NPL ratios. Since the impact of exogenous NPL shocks is mainly channelled by the reduction in profitability associated with higher provisions, this analysis also suggests that liquidating the NPLs at prices that are significantly below their face values could by itself weaken the supply of credit.
Authors’ note: The views expressed here are those of the authors and do not necessarily represent those of the institutions with which they are affiliated.
Accornero, M, P Alessandri, L Carpinelli and A M Sorrentino (2017), “Non-performing loans and the supply of bank credit: Evidence from Italy”, Bank of Italy, Occasional papers no 374.
Acharya, V V and S Steffen (2015), “The ‘greatest’ carry trade ever? Understanding eurozone bank risks”, Journal of Financial Economics 115(2): 215–236.
Aiyar et al. (2016), “A strategy for resolving Europe’s problem loans”, IMF, Staff discussion note.
Altavilla, C, M Pagano and S Simonelli (2017), “Bank exposures and sovereign stress transmission”, Review of Finance 21(6): 2103–2139.
Angelini, P (2018), “Do high levels of NPLs impair banks’ credit allocation?” Bank of Italy, Notes on financial stability and supervision no 11.
Caballero, Hoshi and Kashyap (2008), “Zombie lending and depressed restructuring in Japan”, American Economic Review 98(5): 1943–1977.
Ciavoliello et al. (2016), “What’s the value of NPLs?” Bank of Italy, Notes on Financial Stability and Supervision no 3.
Ciocchetta et al. (2017), “Bad loan recovery rates”, Bank of Italy, Notes on Financial Stability and Supervision no 7.
Farhi E and J Tirole (2018), “Deadly embrace: Sovereign and financial balance sheets doom loops”, Review of Economic Studies, forthcoming.
Nouy, D (2017) “Regulatory and supervisory responses in Europe to the current financial environment”, speech at the High-level meeting on global and regional supervisory priorities, Basel, 18 October.
Obstfeld, M (2017) Presentation at IMF World Economic Outlook, IMF, October.
Potente, F, R Roca et al. (2017), “Risks and challenges of complex financial instruments: An analysis of SSM banks”, Bank of Italy, Occasional papers no 417.
 For instance, Obstfeld (2017) recently stated that “… the NPLs stock remains high, and NPLs limit new lending, are themselves a source of credit crunch.” Nouy (2017) recently argued that “…NPEs keep banks from providing loans to the economy, which is unfortunate at a time when the economy needs to recover.”
 See here.
 See Ciocchetta et al. (2017). This does not imply that book values are inflated. NPL book values are typically higher than market prices for various reasons; under-provisioning is only one of them. See Ciavoliello et al. (2016).