Abstract
Selected Issues
Financial Deepening in Kosovo
Kosovo’s banking sector is, uniquely among the Western Balkans, healthy and growing. But, after starting from scratch in the years before independence, financial penetration remains relatively shallow. Kosovo needs more lending to support stronger economic growth and convergence with Europe. It also needs better lending that is more focused on productive industries. Credit growth has historically been driven – or constrained – by both supply and demand factors in Kosovo, with supply playing a somewhat stronger role. Healthier balance sheets have allowed supply to meet demand as the latter increased in line with improving macroeconomic conditions. Going forward, meaningful financial deepening could be constrained by funding. The authorities, in turn, should focus on maintaining healthy balance sheets, encouraging diversified loan portfolios, exploring new avenues for bank funding, and further reducing structural bottlenecks to lending.
A. Introduction and Background
1. Recent credit growth has been robust, but is it sufficient to support the stronger economic growth that Kosovo needs? Credit depth in Kosovo remains below most of its peers in the Western Balkans, CESEE more broadly, and non-European countries with similar income levels. This suggests that even the higher credit growth of recent years, in which healthier bank balance sheets allowed credit supply to meet increased demand, may not be sufficient to support the economic growth that Kosovo needs to converge with the rest of Europe.
Private Loan Growth
percent, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimates.Private Loan Growth
percent, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimates.Private Loan Growth
percent, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimates.
Credit Depth
Bank credit to the private sector in CESEE (lhs) and at similar income levels (rhs) in percent of GDP, 2016 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: IFS, Conference Board TED, and Staff calculations.1/ The rhs chart plots Kosovo against other countries across the world with a similar per capita income level (ranging $9,000 to $14,000 in constant 2014 PPP dollars) in 2016 or latest (2015).Credit Depth
Bank credit to the private sector in CESEE (lhs) and at similar income levels (rhs) in percent of GDP, 2016 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: IFS, Conference Board TED, and Staff calculations.1/ The rhs chart plots Kosovo against other countries across the world with a similar per capita income level (ranging $9,000 to $14,000 in constant 2014 PPP dollars) in 2016 or latest (2015).Credit Depth
Bank credit to the private sector in CESEE (lhs) and at similar income levels (rhs) in percent of GDP, 2016 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: IFS, Conference Board TED, and Staff calculations.1/ The rhs chart plots Kosovo against other countries across the world with a similar per capita income level (ranging $9,000 to $14,000 in constant 2014 PPP dollars) in 2016 or latest (2015).2. Kosovo’s banking sector started largely from scratch in the early 2000s. Following the 1998-99 war, there was very little banking activity in Kosovo: The banking system held assets equivalent to 4.3 percent of GDP, largely comprised of balances with other banks. Banks’ outstanding loans at end-2000 totaled €3.3 million. This gradually began to shift, although loans did not comprise a majority of banks’ balance sheets until 2005.
3. Foreign funding inflows and independence sustained a boom over 2005-09. Starting at essentially zero following the war, loans comprised roughly half of bank balance sheets by 2005. At this point, credit growth settled into a 20-40 percent y/y range over through 2009 and financial depth nearly doubled from 17 to 32 percent. The boom was driven by postwar reconstruction, pent-up demand, and, later, exuberance during the run-up to and aftermath of independence in 2008. It was facilitated by the appearance of and aggressive targeting by European banking subsidiaries, in line with strong foreign funding leveraging elsewhere in the Western Balkans and Central, Eastern, and Southeastern Europe (CESEE) during this period.
Bank Credit Outstanding
percent 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates.1/ 2001-03 growth = 682, 234, and 169 percent y/y, respectively.Bank Credit Outstanding
percent 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates.1/ 2001-03 growth = 682, 234, and 169 percent y/y, respectively.Bank Credit Outstanding
percent 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates.1/ 2001-03 growth = 682, 234, and 169 percent y/y, respectively.
Share of Assets
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Share of Assets
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Share of Assets
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
4. Kosovo did not face a post-boom bust like many of its neighbors. Much of CESEE, including some banking systems in the Western Balkans, faced strong deleveraging pressures and a collapse in credit following the global financial crisis. Kosovo – which had experienced less pre-crisis leveraging than its peers and whose macroeconomy was largely unaffected by the global crisis given strong donor inflows and limited international trade and financial linkages – did not. Credit growth began to ease significantly in mid-2008 but still averaged 14.5 percent y/y over 2009-11.
Change in Foreign Bank Funding 1/
(percent of GDP)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: BIS, IFS, CBK1/ Change from pre-crisis peak to post-crisis trough each country country. Funding to all sectors. Kosovo estimated through 2015, others 2016Change in Foreign Bank Funding 1/
(percent of GDP)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: BIS, IFS, CBK1/ Change from pre-crisis peak to post-crisis trough each country country. Funding to all sectors. Kosovo estimated through 2015, others 2016Change in Foreign Bank Funding 1/
(percent of GDP)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: BIS, IFS, CBK1/ Change from pre-crisis peak to post-crisis trough each country country. Funding to all sectors. Kosovo estimated through 2015, others 20165. However, financial deepening stalled beginning in 2012. This was likely driven by a combination of slowing economic growth, a more conservative stance toward the region by foreign subsidiaries (which, post-independence, have controlled 80-90 percent of banking system assets) as the euro area crisis deepened, and declining asset quality as loans made during the boom years began to season. These factors were exacerbated by structural bottlenecks to lending in Kosovo related to inefficient courts and collections processes that meant very large collateral requirements, prolonged asset recovery procedures, and high interest rates. Credit growth to the private sector began easing again in mid-2012, and averaged only 3.7 percent y/y from October 2012 to October 2014. Over this period, credit-to-GDP was correspondingly flat.
6. Lending growth has returned since mid-2015. Coinciding with a dramatic drop in lending rates and interest margins, a strong improvement in asset quality, a return to healthy macroeconomic growth, and structural reforms, credit growth has averaged 8.4 percent y/y since June 2015, with Kosovo’s credit-to-GDP rising by 2 percentage points of GDP to 36.8 percent in 2016.
7. However, while a return to credit growth is welcome, the composition of new lending is not ideal. The banking system’s loan portfolio is comprised of roughly two-thirds corporate loans and one-third household lending. The current credit upcycle has seen strong growth in lending to both sectors, especially households. However, the composition of lending within those sectors is not fully geared toward addressing the structural needs of Kosovo’s economy: Most household credit is geared toward consumer goods (new mortgage loans are growing strongly but the market is nascent). Corporate loan growth has been driven largely by lending to the wholesale/retail trade and services sectors and, to a much lesser extent, the manufacturing and mining sectors, in line with the historic trend. A more robust lending boom would entail lending toward more productive activity that would support the much-needed development of productive, export-oriented industries in Kosovo.
Credit Growth
y/y percent
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBKCredit Growth
y/y percent
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBKCredit Growth
y/y percent
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK
Share of Loans
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Share of Loans
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Share of Loans
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
8. The rest of the paper is comprised as follows: Section II will analyze the demand- and supply-side drivers - and possible constraints - to bank lending in Kosovo. Section III will examine in more detail whether existing levels of bank lending are sufficient relative to fundamentals, or whether a credit gap exists in Kosovo. Section IV will discuss the prospects for further financial deepening going forward. Section V will present policy recommendations that follows from this analysis.
B. Drivers of–and Constraints On–Credit Growth
I. The Current Picture
9. The recent return of lending appears to be driven by both supply and demand. The steady recovery in credit growth since 2015 is a welcome development given the relatively low levels of credit penetration in Kosovo. The uptick has coincided with a confluence of positive developments on both the demand and supply side:
Demand-side developments include:
A favorable macroeconomic environment GDP growth averaged 4.5 percent per year during 2005-11 but dropped to 2.5 percent per year while credit stagnated over 201Economic growth since 2015 has been slightly below that of the boom period but well above the years of stagnation.
A sharp drop in lending rates: Average lending rates hovered around 15 percent until mid-2011, when they began a gradual decline of about 200 basis points over three years (accompanied by a steady slowdown in credit growth). A much sharper decline in lending rates began at about the time that credit growth bottomed out in early 2014. Since then, the average lending rate has fallen by roughly 500 basis points. The sharpest drop, of 390 basis points, occurred between October 2014 and April 2015 and coincided with the start of the current credit upcycle.
Lending Rate
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBKLending Rate
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBKLending Rate
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBKSupply-side developments include:
Improved bank balance sheets. Since lending began in earnest in the mid-2000s, the Kosovo banking sector has, in aggregate, been healthy relative to its Western Balkan and CESEE peers. Banks have generally been well-capitalized and liquid and, aside from some weak performance during the period of lending stagnation, profitable. Asset quality suffered as boom-era loans seasoned, with the NPL ratio doubling to a peak of 8.8 percent in early 2014. However, a combination of increased provisioning, targeted writeoffs, and strong credit growth has halved that ratio since, leaving banks more ready and able to lend. At the same time, profits have returned and funding costs have fallen for Kosovo banks.
Lower risk aversion and search for yield: EU-based banks, which account for 64 percent of system assets in Kosovo, pulled back from CESEE and the Western Balkans following the euro area crisis, partially driving the 2012-14 credit slowdown in Kosovo. These banks are once again lending in Kosovo. While they still appear to be taking a relatively more conservative stance than their non-EU and domestic peers, a somewhat reduced level of risk aversion and a search for yield in the context of a low-interest rate environment in Europe has meant a move back toward lending and increased competition among banks in Kosovo.
Financial sector reforms: The authorities have reduced a number of structural bottlenecks and made improvements to the CBK’s supervisory/regulatory infrastructure in recent years, making banks more confident and willing to lend. These include:
judicial reforms that make collateral recovery more reliable, such as a dramatic reduction in the backlog of cases in the court system; the introduction of a system of private enforcement agents to collect on seized assets; and a unique account registry to facilitate the garnishment of accounts (see Box 1 for further detail);
the adoption and refinement of risk-based supervision at the central bank;
the adoption of a best-practice emergency liquidity assistance framework, critical in a unilaterally euroized country without a lender of last resort function;
the establishment of a deposit insurance fund and subsequent steps to ensure adequate funding to cover the banking sector’s deposit base;
and the adoption of a macroprudential policy framework, also important for a unilaterally euroized country without a monetary policy.
Balance Sheet Health
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimatesBalance Sheet Health
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimatesBalance Sheet Health
(percent)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimates
EU-Based Banks Return to Lending
percent, y/y 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: Bank financial statements, CBK, staff estimates.1/ Aggregate of Kosovo’s three largest banks by assets, representing 64 percent of system assets.EU-Based Banks Return to Lending
percent, y/y 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: Bank financial statements, CBK, staff estimates.1/ Aggregate of Kosovo’s three largest banks by assets, representing 64 percent of system assets.EU-Based Banks Return to Lending
percent, y/y 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: Bank financial statements, CBK, staff estimates.1/ Aggregate of Kosovo’s three largest banks by assets, representing 64 percent of system assets.Improving Court Enforcement Procedures in Kosovo
Kosovo’s inefficient court system historically hindered bank lending and growth. Kosovo’s courts have historically suffered from large case backlogs due to low institutional capacity, weak processes and management, and poor alignment of incentives in handling and resolving caseloads. As recently as July 2014, Kosovo’s courts had an outstanding backlog of over 145,000 open cases (of a population of 1.8 million). In turn, as creditors could not efficiently obtain and enforce judgments against debts, they required more collateral and higher lending rates, which deterred bank lending. Lending rates had hovered well over 10 percent and financial depth, at 35 percent of GDP, is still below what Kosovo needs to support stronger economic growth. While other factors were also at play, weak contract enforcement was clearly a major one.
The authorities finally decided to confront the problem. Their efforts, with strong operational help from USAID, have focused on reducing the courts’ backlog of unenforced judgments and improving enforcement procedures. Specifically, they have:
Introduced in 2014 a system of Private Enforcement Agents (PEAs) that offer creditors a more effective route toward enforcing court judgments, recovering assets, and reducing the burden on courts. The new PEAs are certified by the Ministry of Justice (MoJ) through a rigorous examination process.
Made enforcement actions more effective by establishing a unique and centralized registry of bank account holders at the Central Bank of Kosovo (CBK). The registry, which is accessible by PEAs, enables the garnishment of banks accounts, a faster and more reliable way for creditors to collect after a judgment.
Introduced new resources and more efficient methods for case resolution procedures at the courts, including the embedding of CLE backlog reduction officers.
These efforts were met with some resistance from some quarters – for instance, a public initially skeptical of the new PEA system and entrenched elements of the judicial bureaucracy. But the authorities’ commitment has meant that these efforts have maintained momentum through the process. Steps to improve case resolution at the courts have resulted in the closure of over 91,000 court-based enforcement cases since May 2013. In parallel, PEAs have resolved, between May 2014 and December 2015, 4,809 cases and recovered €51.5 million (equivalent to 0.9 percent of GDP); unofficial estimates suggest that the amount recovered has doubled since then. Relatedly, the banking community sees progress in contract enforcement as an important factor behind the sharp recent decline in lending rates.
Court Backlog Clearance Under USAID Program
cumulative
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: USAID/CLECourt Backlog Clearance Under USAID Program
cumulative
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: USAID/CLECourt Backlog Clearance Under USAID Program
cumulative
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: USAID/CLEII. Empirical Analysis
10. The analysis seeks to identify the role of supply and demand in credit growth in Kosovo. To do so, we use a disequilibrium model based on Everaert et al (2015) that assumes that credit demand and supply are influenced by factors in addition to prices (interest rates). Credit demand and credit supply are estimated separately and actual credit growth is assumed to be the lower of the two. This allows us to view the evolution of credit supply and demand growth over the sample period and provides insight into what was driving or constraining credit growth. The model is summarized as:
Estimations for
11. All variables are statistically significant and have the expected signs, with the exception of interest rate margin on the demand side, which is positive – this could be due to the supply effect dominating this variable. The model fits the historical data trend fairly well, with the exception of the outlying variability of the 2007-10 credit boom. The model captures the directional change but, given this large variability, not the magnitude of credit supply and demand during the boom. The model successfully accounts for the magnitude of credit supply and demand from 2011 onward. As a robustness check, when re-running the regressions on a smaller sample of data that begins in June 2009, after the credit boom, some explanatory variables lose their significance but the fit of the model from mid-2009 onward is as good or better. (See Appendix I for regression results and fitted charts.)
12. The model suggests a fairly even contribution of supply and demand to credit growth from mid-2009 onward, with supply factors moderately more important during both the 2011-13 slowdown and the recovery after 2014. Credit demand has increased over the past three years, and banks were healthy enough such that supply did not constrain this surge in demand. Credit growth, in turn, has surged.
13. Following the end of the credit boom, the model helps to explain the factors behind the two distinct periods of post-boom credit growth in Kosovo:
Slowdown, 2011-13: A gradual slowdown of supply began in 2011, with demand subsequently slowing in 2012, and both bottoming out in late 2013/early 2014 (in line with actual credit growth over this period). The supply slowdown appears to have been driven by a sharp decline in asset quality and interest rate margins that began in late 2010/early 2011 and a sharp drop in profitability beginning in 2012. Economic growth and private debt also deteriorated during this period, although the model suggests that credit demand, while declining, exceeded supply and so supply was the dominant factor in the credit growth slowdown.
Recovery, 2014-present: After bottoming out at nearly zero growth in early 2014, lending has gradually but steadily picked up since. Both credit supply and demand have recovered, although supply earlier than demand: While credit supply benefited from significantly improved balance sheets from mid-to-late 2014 onward (which outweighed a sharp decline in margins starting in mid-2014), full recovery in credit demand came later as the macroeconomic environment remained weak through 2014 and a strong reduction in private debt did not begin until early 2015. Given the strong balance sheet cleanup that occurred leading up to this, banks have been able to meet the recovery in demand with ample supply.
Credit Supply and Demand
private credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimatesCredit Supply and Demand
private credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimatesCredit Supply and Demand
private credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimates
Excess Supply
private credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Excess Supply
private credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Excess Supply
private credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
C. Does Kosovo Need More Lending/Financial Deepening to Converge?
14. Low levels of financial intermediation may hamper growth and convergence. By efficiently channeling savings to productive investment and by improving the distribution of capital and risks across the economy, finance positively contributes to growth (for a literature review, see Levine, 2004). Many studies confirm this positive effect (e.g., King and Levine, 1993; Levine and Zervos, 1998; Beck and others, 2000; Cecchetti and others, 2011), even though its magnitude varies across regions, countries and income levels (Barajas and others, 2013). Recent studies find that the relationship between financial depth and growth is nonlinear: beyond a certain threshold, further financial deepening actually harms growth (e.g., Arcand and others, 2012; Cecchetti and Kharroubi, 2012; Aizenman and others, 2015; Sahay and others, 2015). However, at low levels of financial development, the growth dividend of financial deepening can be substantial. In a similar vein, a low level of financial intermediation may constrain growth through a low allocative efficiency.
15. Private credit in Kosovo is low compared to its peers in the region. At about 37 percent of GDP, domestic bank credit to the private sector in Kosovo is 8 percentage points below the Western Balkan countries’ average, which itself is already low compared to other regions in emerging Europe (Figure 1). When looking at a broader measure of private credit that includes cross-border lending and non-bank credit, Kosovo’s private credit depth of 56 percent of GDP is still 10 percentage points below the regional average. While in part reflecting its low per capita income level, Kosovo’s credit depth is also low compared to other countries with a similar per capita income level.
Kosovo: Credit Depth Compared to Peers and Fundamentals
Bank credit to the private sector in percent of GDP, 2016 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: IFS, Conference Board TED, WEO, BIS, WB WDI, and Staff calculations.1/ The data plotted in the top right chart uses a wider measure of private credit that includes non-bank and cross-border credit. The bottom left chart plots Kosovo and other Western Balkan countries (WBS) against other countries across the world with a similar per capita income level (ranging from $9,000 to $14,000 in constant 2014 PPP dollars). The bottom right chart plots the deviation of private credit from its fundamentals-implied level in percent of GDP. For the estimation of the fundamentals-implied level and related credit gap, see the Appendix.Kosovo: Credit Depth Compared to Peers and Fundamentals
Bank credit to the private sector in percent of GDP, 2016 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: IFS, Conference Board TED, WEO, BIS, WB WDI, and Staff calculations.1/ The data plotted in the top right chart uses a wider measure of private credit that includes non-bank and cross-border credit. The bottom left chart plots Kosovo and other Western Balkan countries (WBS) against other countries across the world with a similar per capita income level (ranging from $9,000 to $14,000 in constant 2014 PPP dollars). The bottom right chart plots the deviation of private credit from its fundamentals-implied level in percent of GDP. For the estimation of the fundamentals-implied level and related credit gap, see the Appendix.Kosovo: Credit Depth Compared to Peers and Fundamentals
Bank credit to the private sector in percent of GDP, 2016 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: IFS, Conference Board TED, WEO, BIS, WB WDI, and Staff calculations.1/ The data plotted in the top right chart uses a wider measure of private credit that includes non-bank and cross-border credit. The bottom left chart plots Kosovo and other Western Balkan countries (WBS) against other countries across the world with a similar per capita income level (ranging from $9,000 to $14,000 in constant 2014 PPP dollars). The bottom right chart plots the deviation of private credit from its fundamentals-implied level in percent of GDP. For the estimation of the fundamentals-implied level and related credit gap, see the Appendix.16. Private credit is also below the level predicted by fundamentals. Following IMF (2015), per capita private credit is estimated as a reduced form function of supply and demand components. The fundamental determinants of private credit include per capita GDP and the interest rate on private debt, while country-specific effects control for cross-country differences in initial conditions such as institutional quality, governance, etc. (see Appendix II). The results show that private credit is some 5-6 percentage points of GDP below the level predicted by fundamentals (Figure 1). This finding is robust for different specifications and sample sizes.2
17. These findings point to significant scope for further credit growth that may raise output growth and accelerate convergence, without the risk of financial sector instability. Closing the credit gap in the medium-term would require credit growth to significantly outpace nominal GDP growth for a sustained period of time: baseline projections for nominal GDP imply that it would have to average 9 percent per year to close the gap in the next five years. While rapid credit growth is often associated with a build-up of financial imbalances, Kosovo’s low level of private credit relative to its fundamental determinants, coupled with ample liquidity in the banking sector, provides room for such an expansion in lending that can unlock higher output growth and accelerate convergence to income levels seen in Western Europe.
D. Can Banks Provide the Additional Funding/Deepening that Kosovo Needs While Maintaining Financial Stability?
18. Banks are fairly well-positioned to continue lending. As discussed in section II, from a supply perspective, balance sheets are healthy and improving, funding liquidity is ample, and deposit growth remains healthy. While some structural bottlenecks remain, the authorities have made and are continuing to make strong progress in minimizing these and further bolstering Kosovo’s financial safety net. From a demand perspective, Kosovo enjoys macroeconomic stability, economic growth is stronger than Western Balkan peers, and a similar level of growth is projected to continue into the medium-term.
19. However, prospects for sound financial deepening based on recent credit trends are mixed. Assuming that the rapid pace of credit growth seen since early 2016 continues into the long term and taking into account current GDP projections, Kosovo would experience an increase of its credit-to-GDP ratio by 16 percentage points, to 52 percent, by 2026. This would represent a significant increase that would more than close the credit gap discussed in section III (although this ratio would still be below the New Member States average and several Western Balkans peers in 2016, much less 2026). As noted above, the existing credit gap and ample liquidity in the banking system suggests that the banking sector could maintain this rapid pace of credit growth (averaging 9.5 percent y/y since January 2016) for some time, even with medium-term nominal GDP growth projected to be only 6.0 percent y/y. However, if credit growth were to ease to a more moderate pace that is roughly in line with projected medium-term nominal GDP growth – for instance, the 2012-16 average annual growth rate of 5.6 percent y/y – there would not be any financial deepening at all. Moreover, even if banks were able to maintain the recent higher rate of credit growth over the long term, they could eventually run into funding issues: Assuming recent rates of deposit growth in parallel with recent rates of credit growth, the higher-credit growth scenario would see the banking sector’s loan-to-deposit ratio reach 100 percent by 2024, a level of funding liquidity that may raise financial stability concerns.
Projected Credit/GDP
percent, based on avg annual credit growth per scenario 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.Projected Credit/GDP
percent, based on avg annual credit growth per scenario 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.Projected Credit/GDP
percent, based on avg annual credit growth per scenario 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.
Projected Loan/Deposit
percent, based on avg annual credit and deposit growth
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimatesProjected Loan/Deposit
percent, based on avg annual credit and deposit growth
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimatesProjected Loan/Deposit
percent, based on avg annual credit and deposit growth
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, staff estimates20. While funding liquidity indicators in Kosovo are healthy, current funding levels and trends could still limit meaningful long-term financial deepening. If we assume that recent rates of deposit growth continue and that all deposits are allocated to lending (such that the loan-to-deposit ratio rises to 100 percent by 2026), then credit/GDP would rise by 12 to 15 percentage points, to 49 to 52 percent, by 2026.3 However, as noted above, reaching this level of credit depth would require credit growth of about 9 percent per year over the next 10 years. As such, reaching this level of depth with a perhaps more sustainable rate of credit growth would require Kosovo banks to raise funding to a greater degree than they have in recent years. Taking a more conservative assumption – in which credit grows at the same rate as deposits, rather than growing at a rate fast enough to reach a 100 percent loan-to-deposit ratio in 2026 – financial deepening would be much more limited, with credit-to-GDP increasing by only 1 to 3 percentage points by 2026.
Change in Credit/GDP and Avg Ann Credit Grth
percent, 2016-26 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ Based on average annual deposit rates and assuming LTD = 100%. GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.Change in Credit/GDP and Avg Ann Credit Grth
percent, 2016-26 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ Based on average annual deposit rates and assuming LTD = 100%. GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.Change in Credit/GDP and Avg Ann Credit Grth
percent, 2016-26 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ Based on average annual deposit rates and assuming LTD = 100%. GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.
Projected Credit/GDP
percent, credit = avg annual deposit growth per scenario 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.Projected Credit/GDP
percent, credit = avg annual deposit growth per scenario 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.Projected Credit/GDP
percent, credit = avg annual deposit growth per scenario 1/
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: CBK, IFS, staff estimates1/ GDP based on staff projections through 2022; 2023-26 GDP based on projected 2022 GDP growth.E. Conclusions and Policy Recommendations
21. The return of bank lending in recent years is encouraging, but Kosovo requires further financial deepening to support growth. Following several years of very weak bank lending, credit growth since 2015 has been far more robust than most of Kosovo’s Western Balkans peers. This has been bolstered by improved bank balance sheets, which have allowed supply to meet rising demand as macroeconomic conditions improved. However, financial depth remains relatively shallow – too shallow to support the level of economic growth that Kosovo needs to meaningfully raise incomes to a level closer to that of the rest of Europe.
22. Lending patterns in Kosovo appear to be fairly evenly driven by both demand and supply factors. Our empirical analysis suggests that credit supply and demand track each other fairly closely, with both important factors in credit growth. Most recently, credit growth has increased in line with demand as macroeconomic conditions have improved but, importantly, with healthy balance sheets and ample liquidity, supply has been able to respond to meet that demand. This is an importance difference between Kosovo and most of its Western Balkans peers, who face both less favorable macroeconomic environments and weaker bank balance sheets.
Growth Since Kosovo’s Independence
(percent, average real GDP growth, 2008-16)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: WEOGrowth Since Kosovo’s Independence
(percent, average real GDP growth, 2008-16)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: WEOGrowth Since Kosovo’s Independence
(percent, average real GDP growth, 2008-16)
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Source: WEO23. Policies, in turn, should focus on both improving the conditions for the demand and supply of credit.
On the demand side, the government should maintain the momentum of the past two years with regard to managing macroeconomic stability, including by continuing to manage sustainable budgets that contain unproductive current spending and create space for growth- and productivity-friendly capital investments. It should also continue to take measures to improve Kosovo’s business environment – for instance, by reducing the administrative burden for businesses by simplifying the legislative and regulatory system; reducing informality by simplifying and improving revenue collection; and further improving infrastructure creating an environment for more productive lending opportunities, thus improving not just the amount of lending but also the quality and composition of bank loans.
On the supply side, the central bank should continue its strong supervisory oversight to ensure healthy bank balance sheets, and also seek ways to further improve its supervisory and regulatory framework, for instance by continuing to develop its macroprudential policy framework. The authorities should also seek to further reduce remaining structural bottlenecks to bank lending, for instance via remaining inefficiencies in the court and collection system and in fully establishing a best-practice cadastre system that can effectively identify and protect property rights.
24. The authorities should also explore ways to help diversify banking funding sources in the long term. Kosovo banks currently have ample levels of funding liquidity. However, current rates of deposit growth may not be sufficient to fund more meaningful financial deepening. A local capital market in which banks could issue bonds would help to diversify bank funding sources, although with capital markets still nascent in Kosovo this would be a longer-term policy goal.
Appendix I. Results and Fit of Disequilibrium Model
Summary of Empirical Results
Dependent variable: Private sector credit growth, y/y
Summary of Empirical Results
Dependent variable: Private sector credit growth, y/y
| Full Sample | (2007- | Post-Boom Sample | ||
|---|---|---|---|---|
| 2017) | (2009-2017) | |||
| Demand | Supply | Demand | Supply | |
| Lending Margin | 1.10** | 4.64*** | 0.067 | 0.587 |
| Import growth | .214*** | .090*** | ||
| NPL ratio (t-1) | −3.73*** | −2.02*** | −1.66*** | −1.60*** |
| Equity/Assets (t-1) | 6.15*** | −.052 | ||
| ROA (t-1) | 2.76** | 1.500* | ||
| Adj R2 | 0.591 | 0.730 | 0.581 | 0.567 |
| Observations | 124 | 124 | 95 | 95 |
Summary of Empirical Results
Dependent variable: Private sector credit growth, y/y
| Full Sample | (2007- | Post-Boom Sample | ||
|---|---|---|---|---|
| 2017) | (2009-2017) | |||
| Demand | Supply | Demand | Supply | |
| Lending Margin | 1.10** | 4.64*** | 0.067 | 0.587 |
| Import growth | .214*** | .090*** | ||
| NPL ratio (t-1) | −3.73*** | −2.02*** | −1.66*** | −1.60*** |
| Equity/Assets (t-1) | 6.15*** | −.052 | ||
| ROA (t-1) | 2.76** | 1.500* | ||
| Adj R2 | 0.591 | 0.730 | 0.581 | 0.567 |
| Observations | 124 | 124 | 95 | 95 |
Fit of Model
full sample, credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Fit of Model
full sample, credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Fit of Model
full sample, credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Fit of Model
post-boomsample, credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Fit of Model
post-boomsample, credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Fit of Model
post-boomsample, credit growth, y/y
Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A003
Appendix II. Estimating Fundamentals—Consistent Levels of Credit
1. The long-run relationship between private sector credit and its main determinants is estimated for 34 European countries during 1995–2016. In a stylized, reduced-form model, private sector credit is driven by per capita income that positively affects both credit demand and supply, and the nominal interest rate on private debt, which has a negative effect on demand and a positive effect on supply. The model also includes country-specific constants:
Rt – nominal interest rate on private sector debt;1
t – country index, t – time index.
Private sector debt is composed of domestic bank credit to the non-financial private sector (IFS) and private external debt liabilities (WEO). Unless otherwise mentioned, the data source of the other series is WEO. All series are time demeaned by subtracting the mean across all countries in a given period from the individual country values.2 Regression results are presented in Table 1. The preferred specification is the Arellano-Bond dynamic-panel system GMM (GMM-SYS). The coefficients of real per capita income and nominal interest rate are sizable, and their signs are consistent with theoretical priors.
2. To arrive at fundamentals-consistent private sector credit estimates, country- and time-specific effects are incorporated. Based on GMM-SYS regression results, the long-run relationship between private sector debt and its fundamentals is:
lowercase variables are expressed in natural logarithm of per capita quantities in thousands of 2005 purchasing power parity U.S. dollars; asterisk indicates long-run value. Country-specific effects are included to ensure that the actual series and their fundamentals-consistent counterparts have the same means for each country in the sample, and reflects the assumption that CESEE countries may not converge to a common equilibrium path for private sector credit from different starting points. Common time effects are included, reflecting the assumption that the dynamics of fundamentals have the same impact on the “equilibrium” debt burdens, whether or not they are driven by common time effects or country idiosyncratic factors. Credit gaps are then calculated as the deviation of actual private sector credit from its fundamentals-consistent level.
Determinants of Real per Capita Private Sector Debt in Europe
Hansen test of overidentifying restrictions (whether the instruments, as a group, appear exogenous).
Wooldridge test for autocorrelation in panel data (H0: no first-order autocorrelation).
Test of (n-th) order serial correlation in regression residuals in first differences, N(0,1). Null hypothesis is no autocorrelation.
F-test that all fixed effects are jointly zero.
Instruments for (1) first differences equation: L(2/3).(l_crdprs_ppp_r_pc_dt l_gdp_ppp_r_pc_dt int_rat_dt); and (2) levels equation: DL.(l_crdprs_ppp_r_pc_dt l_gdp_ppp_r_pc_dt int_rat_dt), using the first 50 principal components of the GMM-style instruments.
Determinants of Real per Capita Private Sector Debt in Europe
| Dependent variable | Log of per capita private sector debt in thousand 2005 PPP USD | ||||
|---|---|---|---|---|---|
| Regression model | (1) | (2) | |||
| Estimator | OLS | FE | OLS | FE | GMM-SYS 5 |
| Lagged dependent variable | … | … | 0.90 | 0.76 | 0.64 |
| (0.012)*** | (0.040)*** | (0.101)*** | |||
| Log of per capita GDP in thousand 2005 PPP USD | 1.57 | 1.81 | 0.10 | 0.38 | 0.58 |
| (0.035)*** | (0.302)*** | (0.019)*** | (0.084)*** | (0.252)** | |
| Interest rate (fraction) | −2.21 | −0.71 | −0.44 | −0.51 | −0.92 |
| (0.225)*** | (0.485) | (0.111)*** | (0.141)*** | (0.240)*** | |
| Common intercept | −0.05 | −0.07 | 0.02 | 0.00 | −0.03 |
| (0.017)*** | (0.031)** | (0.004)*** | (0.006) | (0.034) | |
| Country-specific effects | No | Yes | No | Yes | Yes |
| Observations | 619 | 598 | |||
| Number of countries | 34 | 34 | |||
| Adjusted R-squared | 0.90 | 0.89 | 0.99 | 0.99 | … |
| Within adjusted R-squared | … | 0.49 | … | 0.89 | … |
| Chi2 (54) 1 | … | … | … | … | 23.17 |
| F(2,33) 2 | … | 50.47*** | … | … | … |
| AR(1) 3 | … | … | … | … | −2.87*** |
| AR(2) 3 | … | … | … | … | 0.54 |
Hansen test of overidentifying restrictions (whether the instruments, as a group, appear exogenous).
Wooldridge test for autocorrelation in panel data (H0: no first-order autocorrelation).
Test of (n-th) order serial correlation in regression residuals in first differences, N(0,1). Null hypothesis is no autocorrelation.
F-test that all fixed effects are jointly zero.
Instruments for (1) first differences equation: L(2/3).(l_crdprs_ppp_r_pc_dt l_gdp_ppp_r_pc_dt int_rat_dt); and (2) levels equation: DL.(l_crdprs_ppp_r_pc_dt l_gdp_ppp_r_pc_dt int_rat_dt), using the first 50 principal components of the GMM-style instruments.
Determinants of Real per Capita Private Sector Debt in Europe
| Dependent variable | Log of per capita private sector debt in thousand 2005 PPP USD | ||||
|---|---|---|---|---|---|
| Regression model | (1) | (2) | |||
| Estimator | OLS | FE | OLS | FE | GMM-SYS 5 |
| Lagged dependent variable | … | … | 0.90 | 0.76 | 0.64 |
| (0.012)*** | (0.040)*** | (0.101)*** | |||
| Log of per capita GDP in thousand 2005 PPP USD | 1.57 | 1.81 | 0.10 | 0.38 | 0.58 |
| (0.035)*** | (0.302)*** | (0.019)*** | (0.084)*** | (0.252)** | |
| Interest rate (fraction) | −2.21 | −0.71 | −0.44 | −0.51 | −0.92 |
| (0.225)*** | (0.485) | (0.111)*** | (0.141)*** | (0.240)*** | |
| Common intercept | −0.05 | −0.07 | 0.02 | 0.00 | −0.03 |
| (0.017)*** | (0.031)** | (0.004)*** | (0.006) | (0.034) | |
| Country-specific effects | No | Yes | No | Yes | Yes |
| Observations | 619 | 598 | |||
| Number of countries | 34 | 34 | |||
| Adjusted R-squared | 0.90 | 0.89 | 0.99 | 0.99 | … |
| Within adjusted R-squared | … | 0.49 | … | 0.89 | … |
| Chi2 (54) 1 | … | … | … | … | 23.17 |
| F(2,33) 2 | … | 50.47*** | … | … | … |
| AR(1) 3 | … | … | … | … | −2.87*** |
| AR(2) 3 | … | … | … | … | 0.54 |
Hansen test of overidentifying restrictions (whether the instruments, as a group, appear exogenous).
Wooldridge test for autocorrelation in panel data (H0: no first-order autocorrelation).
Test of (n-th) order serial correlation in regression residuals in first differences, N(0,1). Null hypothesis is no autocorrelation.
F-test that all fixed effects are jointly zero.
Instruments for (1) first differences equation: L(2/3).(l_crdprs_ppp_r_pc_dt l_gdp_ppp_r_pc_dt int_rat_dt); and (2) levels equation: DL.(l_crdprs_ppp_r_pc_dt l_gdp_ppp_r_pc_dt int_rat_dt), using the first 50 principal components of the GMM-style instruments.
References
Aizenman, J., Y. Jinjarak, and D. Park (2015), “Financial Development and Output Growth in Developing Asia and Latin America: A Comparative Sectoral Analysis”, NBER working paper 20917
Arcand, J., E. Berkes, and U. Panizza (2012), “Too much Finance?”, IMF working paper 12/161
Barajas, A., R. Chami, and S. Yousefi (2013), “The Finance and Growth Nexus Re-examined: Do All Countries Benefit Equally?”, IMF working paper 13/130
Beck, T., A. Demiguc-Kunt, and R. Levine (2000), “A New Database on Financial Development and Structure”, World Bank Economic Review, 14, pp. 597–605
Beck, T., R. Levine, N. Loayza (2000), “Finance and the Sources of Growth”, Journal of Financial Economics, 58, pp. 261–300
Cecchetti, S., M. Mohanty, and F. Zampolli (2011), “The Real Effect of Debt”, BIS working paper 352
Cecchetti, S. and E. Kharroubi (2015), “Why Does Financial Sector Growth Crowd Out Real Economic Growth?”, BIS working paper 490
Everaert, G., et al (2015), “Does Supply or Demand Drive the Credit Cycle? Evidence from Central, Eastern, and Southeastern Europe”, IMF working paper 15/15
International Monetary Fund (2015), “Central, Eastern, and Southeastern Europe: Mind the Credit Gap”, EUR Regional Economic Issues, May 2015
King, R. and R. Levine (1993), “Finance and Growth: Schumpeter Might Be Right”, Quarterly Journal of Economics, 108(3), pp. 717–37
Levine, R. (2004), “Finance and Growth: Theory and Evidence”, NBER working paper 10766
Levine, R. and S. Zervos (1998), “Stock Markets, Banks, and Economic Growth”, American Economic Review, 88, pp. 537–58
Sahay, R. M. Cihak, P. N’Diaye, A. Barajas, R. Bi, D. Ayala, Y. Gao, A. Kyobe, L. Nguyen, C. Saborowski, K. Svirydzenka, and S. Yousefi (2015), “Rethinking Financial Deepening: Stability and Growth in Emerging Markets”, IMF Staff Discussion Note 15/08
Estimates via maximum likelihood estimation were also tested and did not qualitatively change the results.
It should be noted that the data for Kosovo do not span a full credit cycle, implying that the estimated “equilibrium” level of private credit may be overestimated. When using a more balanced sample that has an equal number of observations before and after the peak in private credit (2002-2016), the credit gap for Kosovo in fact widens somewhat (while narrowing for the other countries plotted in Figure 1, bottom right chart).
In reality the final credit-to-GDP ratio would likely be even lower than this as some funding raised by banks would be allocated to investing in government securities rather than loans.
For EU countries, the implicit interest rate is calculated using sectoral accounts data as the ratio of interest payments (including financial intermediation services indirectly measured) over the average of the beginning and end-period combined stock of debt of firms and households. For other countries, data are mostly for the lending rate, published in the IMF’s IFS database, with gaps in country coverage filled with data for the short-term interest rate published in the OECD’s Economic Outlook database and from national data sources.
This removes nuisance cross-sectional dependency that creates size distortions and makes inference based on two-stage generalized method of moments (GMM) estimates unreliable (Roodman, 2009).
