## _wp11285 - 3. Policy Simulations, the year 2009

## Source details

**Canonical URL:** [_wp11285 - 3. Policy Simulations, the year 2009](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11285.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11285.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11285.pdf.json)

---

### Introduction and research question
- Research question: effectiveness of countercyclical credit subsidies in emerging markets, using a Serbian policy experiment in early 2009.
- Contextual points:
  - IMF (2010) found countries that kept external pre-crisis vulnerabilities relatively low tended to recover faster.
  - Debate over design, effectiveness, and effects of prudential and capital control policies (Ostry et al., 2010; Stiglitz, 2003).
  - Calvo (2010) conjectured credit subsidies could attenuate credit conditions for small- and medium-sized enterprises during a bust.
- Key preliminary conclusion:
  - Model-based estimates suggest that without credit subsidies Serbia’s recession could have been deeper by about 1–2 percentage points of GDP growth.
- Serbian experiment specificities (affecting replicability):
  - Unexpectedly available fiscal resources early 2009 (reallocated from other already budgeted spending).
  - Energetic Ministry of Economy that moved quickly.
  - Account blockages of delinquent debtors threatened corporate liquidity (Serbia-specific feature).
  - Central bank allowed deducting subsidized credit from the reserve requirement base.
  - Prudent supervisory policies prior to the crisis left banks particularly liquid and solvent.
  - High competition in banking: Hirsch-Herfindahl index of 626 (Serbian banking sector described as the least concentrated banking sector in Europe).

### Model of the non-financial sector (framework and calibration)
- Framework:
  - Partial equilibrium model of monopolistic competition with Dixit-Stiglitz technology.
  - Firm production: linear production function q_i = A l_i with A = 0.57.
  - Firms hold own capital k_i ~ G(μ, σ); need working capital and banks lend only above a free cash flow threshold π_t.
  - Households: CES preferences with elasticity of substitution ɵ.
- Key model relations (preserved as in source):
  - Aggregate demand: Q = (∑_{i=1,...,n} q_i^{(ɵ-1)/ɵ})^{ɵ/(ɵ-1)}.
  - Price index: P = (∑_{i=1,...,n} p_i^{(1-ɵ)})^{1/(1-ɵ)}.
  - Residual demand: p_i = (q_i/Q)^{(-1/ɵ)} P.
  - Firms maximize: Max_{q_i} {(1– i –αΔs)p_i q_i – w l_i + (i +αΔs)k_i} subject to p_i and q_i definitions.
  - Price from first derivative: p_i = w/[A(1–1/ɵ)(1– i –αΔs)].
  - Free cash flow: π_i = p_i q_i (1– (i +αΔs) v_i) – w l_i, where v_i = 1– k_i/(p_i q_i).
  - Number of operating firms: n = [1– G(π_t)]M.
- Equilibrium algorithm:
  - Iterate over wage w, number of companies n, and PQ to evaluate p_i, q_i, l_i, π_i, n*, L*, w*, PQ* per relations (7)–(14) and verify convergence.
- Calibration highlights:
  - A = 0.57.
  - Elasticity of substitution ɵ = 2.28, implying gross markup of 22 percent (corresponding to observed average of 21.5 percent in 2008).
  - Labor supply parameter reported as 0.58 and 0.59 in Table 1 entries.
  - Leverage mean: 49.5; standard deviation of leverage: 27 percent.
  - Threshold for receiving credit: exceeding 38 percent of value added.
  - Share of corporate loans that were FX loans: 75 percent (assumed all unhedged).

### Role of the exchange rate and nonlinearities
- Exchange-rate channel:
  - Appreciation during inflows lowers prices, raises quantities, employment, profits, wages, and number of producing companies.
  - Depreciation during outflows/crisis increases debt service on unhedged FX loans, raises prices, forces bank closures of less profitable firms, and reduces real activity.
- Nonlinear effectiveness of credit subsidies:
  - Marginal impact of a percentage point interest-rate subsidy depends on exchange rate shock position relative to the distribution of mark-ups/free cash flows.
  - Example logic: a 1 percentage point subsidy has small effect when depreciation accelerates from 1 to 2 percent, but can have significant effect when depreciation accelerates from 10 to 11 percent if distributional parameters hold (mean mark-up 12 percent, standard deviation 4 percent).

### Serbia 2008–2009 facts used in simulations
- Exchange rate and FX exposure:
  - Until October 2008 RSD/Euro stable; more than 90 percent of bank credit in that currency.
  - RSD depreciation year-on-year in Q4 2008: 9 percent.
  - RSD depreciation year-on-year in 2009: 15 percent (RSD/Euro from 81.3 to 93.1, i.e., 14.5 percent reported elsewhere).
  - Share of unhedged FX loans: 75 percent.
- Interest rates and cost of servicing debt:
  - Weighted average interest rate on stock of loans to non-financial sector: 2008: 14.3 percent; 2009: 12.4 percent (including interest rate subsidy for companies in the amount of 0.9 percentage points).
  - Reported banks sample accounted for 37.2 percent of total banking assets in 2008.
- Labor and GDP facts:
  - Labor productivity: 0.57.
  - Labor supply parameter (calibrated): 0.59.
  - Serbia’s GDP fall in 2009: 3½ percent.
  - Contribution to 2009 recession: Rest of the economy: 1 percentage point; Non-financial sector: 2 percent.
  - Relative procyclicality of the rest of the economy: 80 percent of that in the non-financial sector.

### Credit subsidy scheme details (as implemented early 2009)
- Company liquidity-type loans:
  - Repayment schedule: 12 months.
  - Interest-rate credit subsidy: five percentage points from a capped interest rate of 10.5 percent or flexible Euribor 1M plus eight percent p.a. on FX-denominated loans.
  - In May cap unified to eight percent on all FX-denominated loans (except reprogramming loans cap 9.9 percent).
  - RSD-denominated loans cap: reference rate of the National Bank of Serbia (NBS) less 3.5 percentage points.
  - NBS allowed banks to reduce the base for Required Reserves (RR) for program loans, provided banks do not deleverage.
  - Ceilings from program start:
    - 20 thousand euro or 2 million dinars for entrepreneurs;
    - 2 million euro or 200 million dinars for large companies;
    - Export loans ceilings were double.
- Consumer durable goods loans:
  - Subsidized three-year repayment consumer loans and lump-sum subsidies for locally produced goods; repayment period three years with one-year grace period.
  - Interest rates fixed: 9.5 percent on cars and 11 percent on the rest of Serbian products.
  - Interest rate subsidy: five percent.
  - Buyer incentives: rebate of one thousand euro trade-in cars for locally produced new FIAT Punto cars; subsidy of two thousand euro for locally produced tractors; subsidy of 20 percent for locally produced trucks with additional 10 percent subsidy for trade-in car and 20 percent subsidy for construction machines.
- Subsidized housing loans (mortgages):
  - Limit euro 100 thousand.
  - Limited to first-time home buyers not older than 45 years.
- Subsidized lending totals (footnote details):
  - Subsidized liquidity loans: euro 936.1 million.
  - Subsidized consumer loans: euro 169.4 million.
  - Subsidized investment loans: euro 2.4 million.
  - Subsidized housing loans: euro 21 million.
  - More than 93 percent of total subsidized loans denominated in euro; remainder in RSD.
  - Assumed program efficiency for generating new loans: 25 percent.

### Model fit and baseline simulation results (2008 and 2009 comparisons)
- Model matches key 2008 data exactly or nearly exactly:
  - Number of companies (receiving credit): Data/Model 13,658 / 13,658.
  - Real output per enterprise: 40 / 40.
  - Average number of employees per enterprise: 70 / 69 (mixed cells).
  - Wage bill: 517,900 versus 523,185 (million RSD).
  - Cash flow: 566,130 versus 560,845 (million RSD).
  - Number of employees: 940,903 versus 954,563.
  - Price level: 2.02 versus 2.01 (index).
- 2009 shocks used in simulation:
  - RSD depreciation ~15 percent.
  - Customer lending interest rate decreased by nearly 2.5 percentage points.
  - Sector’s nominal demand increased by 1.6 percent (mainly due to consumer credit subsidy schemes).
- Observed 2009 model-data alignment: simulation closely matches actual 2009 data in Table 2.
- Year-on-year change in share of companies not getting credit: increased from 7 to 12 percent in 2009.

### Subsidy-scenario simulations and quantitative outcomes (Table 3 scenarios)
- Three simulations:
  1. Only corporate subsidies (interest rate subsidy for companies removed).
  2. No subsidies (both corporate and consumer subsidies removed).
  3. Government spending equal to actual costs of subsidies (spending directed to non-financial sector).
- Selected scenario outcomes (preserved exactly as in source):

  Scenario 1 — Only corporate subsidies removed
  - Country’s real output would have declined by 4.2 percent.
  - Closing down additional 1 percent of companies.
  - Effective borrowing costs higher by 0.9 percentage points relative to with-subsidies simulation.
  - Number of employees and wage in non-financial sector: decline by 0.6 percent.
  - Inflation would accelerate by 0.7 percentage points.
  - Sector’s physical output would decline by additional 1 percentage point.
  - Non-financial sector contribution to decline in Serbian real output increases by 0.4 percent to 2.3 percent.
  - Rest of the economy would contribute to fall in output by 1.9 percent (using approximate spillover).

  Scenario 2 — No subsidies
  - 2009 recession would deepen to 5.6 percent (country GDP growth).
  - Subsidized loans in 2009 amounted to euro 1.1 billion, accounting for 10 percent of newly granted loans.
  - Assumed program efficiency: 25 percent.
  - Initial shock to sector demand: 2.5 percent (25 percent from approximately RSD 100 billion).
  - Compared to simulation without interest rate subsidies, additional demand shock led to:
    - Nominal demand decline by 4 percent.
    - Sector physical output, employment, and wage falling by additional 2 percent.
    - Inflation falling by 2.2 percentage points.
  - With spillovers, economy would experience in 2009 a deeper recession by 2.1 percentage points than actually observed.

  Scenario 3 — Government spending equal to subsidy costs
  - Sector output declines less by 0.6 percentage points compared to no-subsidy case.
  - Inflation speeds up by 0.6 percentage points relative to no-subsidy case.
  - Monthly wage and employment increase by 0.6 percentage points relative to no-subsidy case.
  - Physical output of entire economy would fall by 5.2 percent (only 0.4 percentage points less than no-subsidy case).
  - Conclusion: credit subsidies appear more effective than simple government spending in supporting demand in these simulations.

- Selected values from Table 3 (preserved):
  - Number of companies (receiving credit):
    - Consumer and corporate subsidies: 12,871.
    - Only corporate subsidies: 12,724.
    - No subsidies: 12,724.
    - Government spending: 12,724.
  - Share of companies that did not qualify for credit (percent): 12.3; 13.3; 13.3; 13.3 (respectively).
  - Number of employees:
    - Consumer and corporate subsidies: 894,726.
    - Only corporate subsidies: 888,848.
    - No subsidies: 870,432.
    - Government spending: 875,817.
  - Nominal gross value added (million RSD): 1,101,756; 1,101,756; 1,056,574; 1,069,689 (respectively).
  - Annual wage bill (million RSD): 472,315; 466,130; 447,015; 452,563 (respectively).
  - Annual cash flow (million RSD): 629,441; 635,626; 609,559; 617,126 (respectively).
  - Interest rate (weighted average RSD & FX, percent): 12.4; 13.4; 13.4; 13.4 (respectively).
  - Price level change (y-o-y inflation, percent): 6.9; 7.6; 5.4; 6.0 (respectively).
  - Sector's GDP growth rate (percent): -4.9; -5.9; -7.9; -7.3 (respectively).
  - Sector's contribution to country's real GDP growth (percent): -1.9; -2.3; -3.1; -2.8 (respectively).
  - Country's GDP growth (percent): -3.5; -4.2; -5.6; -5.2 (respectively).

### Key findings and policy implications
- Main quantitative conclusion:
  - Estimated counterfactuals suggest subsidy schemes reduced the 2009 recession by an estimated 1–2 percentage points:
    - Actual observed real GDP decline: 3.5 percent in 2009.
    - Counterfactual without any subsidy schemes: real GDP decline would have been 5.6 percent.
  - Non-financial sector would have added 1.2 percentage points to recession; rest of economy would have added 0.9 percentage points in the no-subsidy counterfactual.
  - Nearly 150 additional companies would have been closed; employment rate in non-financial sector would have fallen by an additional 2.7 percent.
- Mechanisms of effectiveness:
  - Enterprise interest-rate credit subsidies reduced borrowing costs, increased free cash flows, eased credit renewal, and reduced loan rejection rates by about one percentage point.
  - Consumer credit subsidies raised nominal demand and indirectly improved firm cash flows.
  - Because much lending was unhedged FX exposure and depreciation amplified exits, subsidies had nonlinear impacts depending on shock severity relative to distributional parameters.
- Comparison with alternative fiscal action:
  - Simulations indicate credit subsidies were more effective than equivalent direct government purchases in supporting demand for the non-financial sector.
- Replicability and caveats:
  - Serbian case relied on a confluence of favorable institutional and timing factors: available fiscal resources, quick administrative action, central bank RR relief, previously prudent supervision leaving banks liquid, and a highly competitive banking sector.
  - Social optimality of countercyclical credit subsidies is not assessed here; longer-term intertemporal growth and redistribution effects require further analysis.

*Source: _wp11285 - 3. Policy Simulations, the year 2009 (PDF chapter content provided).*

### References .............................................................................................................

### _wp11285 - References .............................................................................................................

### References
- References .................................................................................................................................... 19

### Tables
- 1. Descriptive Statistics ...................................................................................... 12
- 2. Model’s Fit ............................................................................................... 15

*https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11285.pdf*

### 3. Policy Simulations, the year 2009...................................................................... 16

### 3. Policy Simulations, the year 2009

### Introduction
- Research question: effectiveness of countercyclical credit subsidies in emerging markets, using a Serbian policy experiment in early 2009.
- Contextual points:
  - IMF (2010) found countries that kept external pre-crisis vulnerabilities relatively low tended to recover faster.
  - Debate exists over design, effectiveness, and effects of prudential and capital control policies (Ostry et al., 2010; Stiglitz, 2003).
  - Calvo (2010) conjectured credit subsidies could attenuate credit conditions for small- and medium-sized enterprises during a bust.
- Key preliminary conclusion:
  - Model-based estimates suggest that without credit subsidies Serbia’s recession could have been deeper by about 1–2 percentage points of GDP growth.
- Specificities making the Serbian experiment unusual and potentially hard to replicate:
  - Unexpectedly available fiscal resources early 2009 (reallocated from other already budgeted spending).
  - Energetic Ministry of Economy that moved quickly.
  - Account blockages of delinquent debtors threatened corporate liquidity (Serbia-specific feature).
  - Central bank allowed deducting subsidized credit from the reserve requirement base.
  - Prudent supervisory policies prior to the crisis left banks particularly liquid and solvent.
  - High competition in banking: Hirsch-Herfindahl index of 626 (Serbian banking sector described as the least concentrated banking sector in Europe).

### The model of a non-financial sector
- Framework:
  - Partial equilibrium model of monopolistic competition with Dixit-Stiglitz technology.
  - Focus on small, medium, and large companies; linear production function q_i = A l_i with A = 0.57.
  - Companies hold own capital k_i ~ G(μ, σ); need working capital and banks lend only above a free cash flow threshold π_t.
  - Households: CES preferences with elasticity of substitution ɵ.
- Key equations and constructs (preserved as in source):
  - Aggregate demand: Q = (∑_{i=1,...,n} q_i^{(ɵ-1)/ɵ})^{ɵ/(ɵ-1)}.
  - Price index: P = (∑_{i=1,...,n} p_i^{(1-ɵ)})^{1/(1-ɵ)}.
  - Residual demand: p_i = (q_i/Q)^{(-1/ɵ)} P.
  - Firms maximize: Max_{q_i} {(1– i –αΔs)p_i q_i – w l_i + (i +αΔs)k_i} subject to p_i and q_i definitions.
  - Price from first derivative: p_i = w/[A(1–1/ɵ)(1– i –αΔs)].
  - Free cash flow: π_i = p_i q_i (1– (i +αΔs) v_i) – w l_i, where v_i = 1– k_i/(p_i q_i).
  - Number of operating firms: n = [1– G(π_t)]M.
- Equilibrium algorithm (iterate over wage w, number of companies n, and PQ):
  - Evaluate p_i, q_i, l_i, π_i, n*, L*, w*, PQ* as per relations (7)–(14) and verify convergence.

### The role of the exchange rate
- Exchange rate amplifies business cycles in emerging markets via FX borrowing:
  - During excessive capital inflows: appreciation lowers prices, raises quantities, employment, profits, wages, and number of producing companies.
  - During outflows/crisis: depreciation increases debt service on unhedged FX loans, raises prices, forces bank closures of less profitable firms, and reduces real activity.
- Empirical magnification: because distributions of free cash flows and mark-ups are bell-shaped and often narrow, marginal exchange rate depreciation can have increasing marginal effects on firm exits.

### The effects of credit subsidies (mechanism)
- Types modeled:
  - Enterprise interest-rate credit subsidies: reduce borrowing costs for firms, increase free cash flows, ease credit renewal.
  - Consumer credit subsidies: increase demand for firms’ products, indirectly improve firm cash flows.
- Nonlinear effectiveness:
  - The marginal impact of a percentage point interest-rate subsidy depends on the position of the exchange rate shock relative to the distribution of mark-ups/free cash flows.
  - Example logic preserved from text: a 1 percentage point subsidy has small effect when depreciation accelerates from 1 to 2 percent, but can have significant effect when depreciation accelerates from 10 to 11 percent if distributional parameters hold (mean mark-up 12 percent, standard deviation 4 percent).

### The rest of the economy (modeling approach)
- Rest of economy comprises micro businesses, banks, and government sector; they are less explicitly modeled.
- Approach: apply observed relative procyclicality of the non-financial sector relative to the rest of the economy (as observed in 2008 and 2009) to map simulated non-financial sector equilibria into broader economy impacts.

### Calibration and simulation for Serbia
- Crisis facts:
  - Until October 2008 the RSD/Euro exchange rate remained stable; more than 90 percent of bank credit was granted in that currency.
  - In Q4 2008 the RSD depreciated by 9 percent year-on-year.
  - In 2009 it lost 15 percent year-on-year.
- Early-2009 policy: government introduced credit subsidies to ease global financial crisis spillovers.
- Credit subsidy scheme key details:
  - Liquidity-type loans per company: repayment schedule 12 months.
  - Interest rate credit subsidy: five percentage points from a capped interest rate of 10.5 percent or flexible Euribor 1M plus eight percent p.a. on FX-denominated loans.
  - In May cap unified to eight percent on all FX-denominated loans (except reprogramming loans cap 9.9 percent).
  - RSD-denominated loans cap: reference rate of the National Bank of Serbia (NBS) less 3.5 percentage points.
  - NBS allowed banks to reduce the base for Required Reserves (RR) for program loans, provided banks do not deleverage.
  - Ceilings from program start:
    - 20 thousand euro or 2 million dinars for entrepreneurs;
    - 2 million euro or 200 million dinars for large companies;
    - export loans ceilings were double.
  - Consumer durable goods loans: subsidized three-year repayment consumer loans and lump-sum subsidies for locally produced goods; repayment period three years with one-year grace period.
    - Interest rates fixed: 9.5 percent on cars and 11 percent on the rest of Serbian products.
    - Interest rate subsidy: five percent.
    - Buyer incentives:
      - Rebate of one thousand euro trade-in cars for locally produced new FIAT Punto cars.
      - Subsidy of two thousand euro for locally produced tractors.
      - Subsidy of 20 percent for locally produced trucks with additional 10 percent subsidy for trade-in car and 20 percent subsidy for construction machines.
  - Subsidized housing loans (mortgages):
    - Limit euro 100 thousand.
    - Limited to first-time home buyers not older than 45 years.
- Model calibration highlights:
  - Labor productivity parameter A = 0.57 (linear production function q_i = A l_i).
  - Elasticity of substitution ɵ = 2.28, implying gross markup of 22 percent (corresponding to observed average of 21.5 percent in 2008).
  - Labor supply parameter reported as 0.58 and 0.59 in Table 1 entries.
  - Sector descriptive statistics (selected, preserved exactly as in source):
    - Sector's nominal GDP growth rate (percent): -3.6; -0.7; 4.9; 1.6 (values appear in table across categories).
    - Sector's real GDP growth rate (percent): -9.7; -7.0; -1.8; -4.8.
    - Number of enterprises: 10,415; 2,675; 568; 13,658 (table by size).
    - Number of employees: 214,136; 281,865; 458,562; 954,563 and alternate year values 200,954; 259,129; 435,751; 895,834.
    - Average number of employees per company: 2110; 5807; 69.9; 2010; 5824; 69.6 (table-presented values).
    - Physical output per employee: 0.54; 0.48; 0.63; 0.57 (and alternate 0.52; 0.48; 0.65; 0.57).
    - Gross markup (percent): 16.7; 20.6; 24.8; 21.5 and alternate 18.3; 20.8; 29.3; 24.0 (table values).
    - Leverage (population of companies, percent): 49.4.
    - Standard deviation of leverage (percent): 27.
    - Cost of credit (percent): 14.9; 12.4.
    - Interest rate (percent): 13.3.
    - Interest rate subsidy (percent): -0.9.
- Empirical calibration rationale:
  - Homogeneous physical output per worker across firm sizes justifies linear-in-labor production and single productivity parameter A = 0.57.
  - Similar gross markups across firm sizes justify imperfect competition with equal markups.
  - Exit of companies during downturn was relatively uniform across firm sizes, supporting model aggregation.

### Key findings and policy implications (as reported)
- Serbia’s credit subsidy scheme apparently helped buffer the economy against a deeper recession in 2009.
- Model-based counterfactual suggests absence of subsidies could have deepened recession by about 1–2 percentage points of GDP growth.
- Social optimality of countercyclical credit subsidies is not assessed in this paper and would require longer-term intertemporal consideration of growth and redistribution effects.
- Replicability caveats: the Serbian case depended on a confluence of favorable institutional and timing factors (listed above).

*Source: _wp11285 - 3. Policy Simulations, the year 2009 (PDF chapter content provided).*

### 2009. Although large companies charge on average slightly higher markups than the other size-

### _wp11285 - 2009. Although large companies charge on average slightly higher markups than the other size-

### Leverage distribution and credit threshold
- Leverage is modeled as continuous and distributed normally.
- Calibrated mean of leverage: 49.5
- Calibrated standard deviation of leverage: 27 percent
- Loan rejection rate in Eastern Europe (survey 2004–05): 7.6 percent
- Threshold for receiving credit (free cash flows required): exceeding 38 percent of value added
- Assumption: threshold unchanged for 2009; credit standards reported by banks remained rather unchanged in 2009
- Number of actual companies in 2008 scaled by credit approval rate implies 14,686 companies would have existed in 2008 if banks granted loans to all applicants

### FX exposure, interest rates, and cost of servicing debt
- Share of corporate loans that were FX loans (NBS statistics): 75 percent
- Assumption: all FX loans were unhedged
- RSD depreciation year-on-year in 2009 against euro: from 81.3 to 93.1 (i.e., 14.5 percent)
- Weighted average interest rate (on RSD and FX) on stock of loans to non-financial sector:
  - 2008: 14.3 percent
  - 2009: 12.4 percent (including interest rate subsidy for companies in the amount of 0.9 percentage points)
- Reported banks sample accounted for 37.2 percent of total banking assets in 2008

### Labor supply and procyclicality
- Labor supply parameter (calibrated): 0.59
- Labor productivity: 0.57
- Serbia’s GDP fall in 2009: 3½ percent
- Contribution to 2009 recession:
  - Rest of the economy: 1 percentage point
  - Non-financial sector: 2 percent
- Relative procyclicality of the rest of the economy: 80 percent of that in the non-financial sector

### Model fit (2008 and 2009 comparisons)
- Model matches 2008 companies’ data:
  - Number of companies (receiving credit): Data/Model 13,658 / 13,658 (exact match)
  - Real output per enterprise: 40 / 40 (model vs data)
  - Average number of employees per enterprise: 70 / 69 (data/model reported as 70 versus 69 in mixed cells)
  - Markup: exact match
- Aggregate comparisons (Model versus Data for 2008):
  - Wage bill: 517,900 versus 523,185 (million RSD)
  - Cash flow: 566,130 versus 560,845 (million RSD)
  - Number of employees: 940,903 versus 954,563
  - Wage: 45,869 versus 45,674
  - Physical output: 536,315 versus 538,506
  - Price level: 2.02 versus 2.01 (index)
- 2009 shocks used in simulation: RSD depreciation ~15 percent, customer lending interest rate decreased by nearly 2.5 percentage points, sector’s nominal demand increased by 1.6 percent (mainly due to consumer credit subsidy schemes)
- Observed 2009 model-data alignment: simulation closely matches actual 2009 data in Table 2
- Year-on-year change in share of companies not getting credit: increased from 7 to 12 percent in 2009

Key 2008–2009 sector indicators (from Table 2; present as in source)
- Companies: Number receiving credit — 2008: 13,658; 2009: 12,872 (Model/Data columns show 13,658 / 12,871 in some entries)
- Share of companies that did not qualify for credit (percent): 7.0 (2008) and 12.3 (2009)
- Real output per enterprise: 40 (2008) and 39–40 (2009 entries)
- Individual product price (index, 100=2000): 2.02 (2008) and 2.16 (2009)
- Number of employees: Data/Model 954,563 / 940,903 (2008) and 895,834 / 894,726 (2009)
- Average number of employees per enterprise: 70 (2008) and 69–70 (2009)
- Nominal gross value added (million RSD): 1,084,030 (2008) and 1,101,756 (2009)
- Price-cost margin (profit margin, percent): 51.7–52.2 (2008) and 56.9–57.1 (2009)
- Average actual profit margin (after debt service, percent): 44.0 (2008) and 44.6 (2009)
- Standard deviation of actual profit margins (percent): 4.00 (2008) and 6.00 (2009)
- Elasticity of substitution: 2.28 (2008 and 2009)
- Annual wage bill (million RSD): 523,185 (2008) and 517,900 (model) / 472,315 (data for 2009)
- Annual cash flow (million RSD): 560,845 (2008) and 566,130 (model) / 629,441 (data for 2009)
- Credit extended by banks (claims on the sector): 534,128 (2008) and 622,199 (2009)
- Interest rate (weighted average RSD & FX, percent): 14.9 (2008) and 12.4 (2009)
- Share of unhedged FX loans (percent): 75 (both years)
- Exchange rate change (y-o-y, percent, RSD/Euro): 0.0 (2008) and 15.0 (2009)
- Price level (GDP deflator, 100 = 2002): 2.01 (2008) and 2.16 (2009)
- Price level change (y-o-y inflation, percent): 6.7 (2008) and 6.9 (2009)
- Sector's GDP (const. prices 2002, million RSD): 538,506 (2008) and 509,994 (2009)
- Sector's GDP growth rate (percent): -4.8 (2008) and -4.9 (2009)
- Country's GDP (current prices, million RSD): 2,722,461 (2008) and 2,815,000 (2009)
- Share of sector's GDP in country's nominal GDP (percent): 40 (2008) and 39 (2009)
- Sector's contribution to country's real GDP growth (percent): -1.9 (both years)

Sectoral outcomes in 2009 described in text:
- Operating companies raised prices by 7 percent
- Price-cost margin increased to 57 percent
- Sector’s physical output declined by 5 percent
- Declining employment mostly due to layoffs from closing companies

### Subsidy scheme simulations (Table 3 scenarios)
- Three simulations:
  1. Only corporate subsidies (interest rate subsidy for companies removed)
  2. No subsidies (both corporate and consumer subsidies removed)
  3. Government spending equal to actual costs of subsidies (spending directed to non-financial sector)

Scenario 1: Without interest rate subsidies for companies (compared to observed outcome)
- Country’s real output would have declined by 4.2 percent
- Absence of interest rate subsidies implies closing down additional 1 percent of companies
- Effective borrowing costs higher by 0.9 percentage points relative to with-subsidies simulation
- Additional impacts:
  - Average number of employees and physical output per employee: no change relative to simulation with subsidies
  - Number of employees and wage in non-financial sector: decline by 0.6 percent
  - Inflation would accelerate by 0.7 percentage points
  - Sector’s physical output would decline by additional 1 percentage point
  - Non-financial sector contribution to decline in Serbian real output increases by 0.4 percent to 2.3 percent
  - Rest of the economy would contribute to fall in output by 1.9 percent (using approximate spillover)

Scenario 2: No subsidies (removing effects of credit subsidies on sector demand)
- 2009 recession would deepen to 5.6 percent (country GDP growth)
- Subsidized loans in 2009 amounted to euro 1.1 billion, accounting for 10 percent of newly granted loans
- Assumed program efficiency: 25 percent
- Initial shock to sector demand: 2.5 percent (25 percent from approximately RSD 100 billion)
- Compared to simulation without interest rate subsidies, this additional demand shock led to:
  - Nominal demand decline by 4 percent
  - Sector physical output, employment, and wage falling by additional 2 percent
  - Inflation falling by 2.2 percentage points
- With spillovers, economy would experience in 2009 a deeper recession by 2.1 percentage points than actually observed

Scenario 3: Government spending equal to subsidy costs (direct purchases from non-financial sector)
- Simulated equilibrium shows:
  - Sector output declines less by 0.6 percentage points compared to no-subsidy case
  - Inflation speeds up by 0.6 percentage points relative to no-subsidy case
  - Monthly wage and employment increase by 0.6 percentage points relative to no-subsidy case
  - Physical output of entire economy would fall by 5.2 percent (only 0.4 percentage points less than no-subsidy case)
- Conclusion: credit subsidies appear more effective than simple government spending in supporting demand in these simulations

Selected values from Table 3 (as presented)
- Number of companies (receiving credit) under:
  - Consumer and corporate subsidies: 12,871
  - Only corporate subsidies: 12,724
  - No subsidies: 12,724
  - Government spending: 12,724
- Share of companies that did not qualify for credit (percent):
  - Consumer and corporate subsidies: 12.3
  - Only corporate subsidies: 13.3
  - No subsidies: 13.3
  - Government spending: 13.3
- Real output per enterprise: 40 (all scenarios) and 39 in some Model columns
- Individual product price (index, 100=2000): 2.16 (consumer & corporate), 2.17 (only corporate), 2.13 (no subsidies), 2.14 (government spending)
- Number of employees:
  - Consumer and corporate subsidies: 894,726
  - Only corporate subsidies: 888,848
  - No subsidies: 870,432
  - Government spending: 875,817
- Average number of employees per enterprise: 70 (consumer & corporate), 70 (only corporate), 68 (no subsidies), 69 (government spending)
- Nominal gross value added (million RSD):
  - Consumer and corporate subsidies: 1,101,756
  - Only corporate subsidies: 1,101,756
  - No subsidies: 1,056,574
  - Government spending: 1,069,689
- Price-cost margin (profit margin, percent): 57.1–57.7 across scenarios
- Annual wage bill (million RSD): 472,315 (consumer & corporate), 466,130 (only corporate), 447,015 (no subsidies), 452,563 (government spending)
- Annual cash flow (million RSD): 629,441; 635,626; 609,559; 617,126 (respectively)
- Interest rate (weighted average RSD & FX, percent): 12.4 (consumer & corporate), 13.4 (only corporate), 13.4 (no subsidies), 13.4 (government spending)
- Price level (GDP deflator, 100 = 2002): 2.16; 2.17; 2.13; 2.14 (respectively)
- Price level change (y-o-y inflation, percent): 6.9; 7.6; 5.4; 6.0 (respectively)
- Sector's GDP (const. prices 2002, million RSD): 509,994; 506,643; 496,146; 499,216 (respectively)
- Sector's GDP growth rate (percent): -4.9; -5.9; -7.9; -7.3 (respectively)
- Sector's contribution to country's real GDP growth (percent): -1.9; -2.3; -3.1; -2.8 (respectively)
- Country's GDP growth (percent): -3.5; -4.2; -5.6; -5.2 (respectively)

Additional details on subsidized lending in 2009 (footnote information)
- Subsidized liquidity loans: euro 936.1 million
- Subsidized consumer loans: euro 169.4 million
- Subsidized investment loans: euro 2.4 million
- More than 93 percent of total subsidized loans denominated in euro; remainder in RSD
- Subsidized housing loans: euro 21 million
- Assumed program efficiency for generating new loans: 25 percent (based on various international evidences cited)

### Conclusions and policy implications
- Serbia introduced credit subsidy programs as countercyclical measures in late 2008–2009, targeting supply (interest rate subsidies to companies, conditional on not laying off employees) and demand (subsidized consumer and mortgage loans).
- Estimated counterfactuals suggest subsidy schemes reduced the 2009 recession by an estimated 1–2 percentage points:
  - Actual observed real GDP decline: 3.5 percent in 2009
  - Counterfactual without any subsidy schemes: real GDP decline would have been 5.6 percent
  - Non-financial sector would have added 1.2 percentage points to recession; rest of economy would have added 0.9 percentage points
  - Nearly 150 additional companies would have been closed; employment rate in non-financial sector would have fallen by an additional 2.7 percent
- Serbian-specific factors increased effectiveness of subsidy schemes relative to direct government spending:
  - Government had incidentally available resources and adjusted schemes in response to developments
  - High competition in banking sector and ample bank liquidity facilitated success in promoting subsidized lending
  - Credit subsidy for companies reduced loan rejection rate by one percentage point
  - Credit subsidies increased aggregate demand more than equivalent direct government purchases in model simulations
- Caveat: it is too early to assess whether the subsidy schemes were socially optimal; the programs entail intertemporal and distributional effects that require further assessment

*Source: IMF calculations and supporting data as presented in the supplied content.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2011/_wp11285.pdf_