## _wp15274 ## Source details **Canonical URL:** [_wp15274](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15274.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15274.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15274.pdf.json) --- ### Overview and motivation - Two entrenched distortions in China’s financial sector: - administratively-controlled deposit interest rate ceilings; and - widespread implicit state guarantees for financial institutions and corporates (particularly SOEs). - Distortions lowered cost of capital and supported high investment but produced misallocation of capital and buildup of credit to low-return activities. - Paper calibrates a heterogeneous-agent general equilibrium model à la Bewley-Aiyagari to Chinese data to simulate effects of: - lifting deposit rate ceilings; and - removing implicit guarantees. ### Key background facts and calibration targets - Deposit interest rate ceilings historically ranged from "0.35 percent" (demand deposits) to "3.3 percent" (1-year time deposits); later allowed up to "1.1 times" higher via a widening “float.” - Average effective deposit rate: between "1-2 percent" even after full liberalization in October 2015. - Yields on alternative saving products (WMPs and internet-based MMFs): "4 to 6 percent"; accounted for about "13 percent" of total bank deposits by mid-2015. - Bank cleanup costs (early 2000s) estimated at more than "20 percent of GDP" by mid-2000s (Ma, 2006). - World Bank Enterprise Survey (2012) probit marginal effects: POE: "0.123***" (and other POE effects "0.117***", "0.119***", "0.109***"); manufacturing: "0.0456***", "0.0667***", "0.0590***"; retail: "0.134***", "0.0844**", "0.0760*"; log(employees): "-0.0390***", "-0.0371***"; log(years of operation): "-0.0144", "-0.0323*"; % of production exported: "-0.120***", "-0.0656*"; growth of sales: "-0.0657**", "-0.0513*"; observations: "2,734", "2,734", "2,514", "2,437". Interpretation: controlling for firm characteristics, a POE is "12 percent" more likely to be credit constrained; 23 percent of all firms in the survey were constrained. - Persistent borrowing-cost differential: spread of "200 bps" between SOE and POE borrowing rates in early to mid 2000s (Ferri and Liu, 2010). ### Model structure (summary) - Time: discrete, annual periods. - Agents: heterogeneous households (workers and entrepreneurs), POEs (private firms), SOEs, banks (state owned, monopolistically competitive), and a government with balanced budget. - POEs face collateral constraint: a_t ≤ k^{ext}_t ≤ λ a_t with collateral parameter λ. - SOEs face no collateral constraints and borrow at r^{l}_{t,soe} ≤ r^{l}_{t,poe}; SOE production: y_{t,soe} = A k_{t,soe}^{α} l_{t,soe}^{1−α}. - Banks: deposit funding at real rate r^d_t; intermediation cost χ; markup μ determines lending rates; implicit guarantee modeled as subsidy κ per unit lent to SOEs. - Two key distortions explicitly modeled: 1. Deposit interest rate ceiling r̄ (binding → multiplier ξ_t > 0). 2. Implicit guarantees to SOEs (subsidy κ). ### Calibration (selected parameters and matched moments) - Preference and technology: - σ = 1.5 (EIS = 0.6) - δ = 0.06 - α = 0.4 - ν = 0.21 - p_e = 0.894 - A = 1 - Banking markup: - baseline μ = 1.1 - Matched moments and implied parameter values: - Nominal deposit interest rate: 330 bps — Model: 330 bps — ̄r = 0.008 - SOEs vs POEs spread: 200 bps — Model: 200 bps — κ = 0.02/1.1 - POEs nominal loan rate: 786 bps — Model: 793 bps — β = 0.892 - Collateral (% of loan): 137% — Model: 137% — λ = 1.73 - Banks overhead cost (% of assets): 1.15% — Model: 1.15% — χ = 0.0115 - VAT rate: 17% — Model: 17% — τ_c = 0.17 - Government consumption (% GDP): 13.5% — Model: 13.5% — g = 0.135 - Labor share SOEs: Data 18.4% — Model 21.5% - Labor held by top 5% private firms: Data 46.0% — Model 46.5% — e_{low} = 0.945 - Pareto shape for entrepreneurial ability: η = 3.63 - Additional calibration notes: - Nominal deposit interest rate set at ceiling as of summer 2014. - Annual inflation rate assumption: 2.5%. - Labor distribution from NBS for 2009. ### Main simulation findings — liberalizing deposit interest rates (section 6.1) - Comparative stationary equilibria: with deposit rate ceiling (̄r) vs without. - Key quantitative outcomes when removing ceiling (baseline calibration, μ = 1.1): - Nominal deposit interest rate rises from ceiling of 330 bps to around 580 bps. - Lending interest rate expected to decline by around 50 bps for both POEs and SOEs. - Capital intensity increases; capital stock boost raises GDP by around 4 percent. - Total TFP falls (reduction in TFP) driven by diminishing marginal returns to capital and additional capital allocated to less productive SOEs enjoying implicit guarantees. - Banking margins and competition: - Movement in deposit and lending rates shrinks banking margins. - Outcomes sensitive to μ: - μ = 1.3 → deposit interest rate increases only to 500 bps. - μ = 1.1 → deposit rate goes up to around 580 bps. - When competition is lower (higher μ), lending rates remain higher and effects on GDP, TFP and capital-output ratio are reduced. - Risks not captured by the model: - Possible bank risk-shifting behavior could change portfolio composition and increase lending interest rates. - Stronger competition can increase incentives for banks to take risk, harming financial stability (cited literature: Vives, 2010; country examples: US in the 1980s, Spain). ### Effects of removing implicit guarantees (with and without deposit liberalization) - Removing implicit guarantees alone: - Leads to more efficient allocation of capital and higher GDP, mainly by reducing SOE role. - Combined reforms (liberalize deposit rates + remove implicit guarantees): - Deposit rate is pushed downward relative to the no-guarantee case. - POEs experience a significant fall in their lending rate. - Capital intensity decreases (less capital available overall). - Labor share of SOEs drops from 22 percent to around 12 percent. - GDP increases by around 1 percent. - Total TFP increases by 3 percent. - Mechanism intuition: - Removing κ shifts SOE demand for capital left (SOEs lose subsidy) → deposit and POE lending rates fall → POEs access more credit → income and savings increase → savings curve shifts right. In calibration, demand-side reduction dominates, net effect is reduced capital intensity but higher TFP. - SOE lending-rate change: - SOEs see a slight rise in their lending interest rates (around 20 bps), substantially less than the removed 200 bps subsidy because general equilibrium feedback lowers r^{l}_{poe} and partially offsets the direct rise. ### Interaction with savings elasticity and other reforms (section 6.3) - Reforms strengthening social safety nets or lifting the one-child policy can shift aggregate savings left; modeled via shock to discount factor β. - Illustrative calibration: shock calibrated so total stock of savings does not change after interest rate liberalization. - Key comparative insights: - If savings are inelastic (no change in total savings), liberalization has smaller GDP impact because capital stock does not expand. - Removing implicit guarantees still increases total TFP even when capital stock is fixed. - The economy’s reaction depends critically on elasticity of savings to interest rate changes. ### Policy implications, risks, and recommendations - Partial reforms (liberalizing deposit rates while implicit guarantees remain) are insufficient and may worsen inefficiency by channeling larger savings into less efficient SOEs and entities with public support. - Removing implicit guarantees is essential to reallocate credit toward more efficient private enterprises and to boost total factor productivity. - Political and financial-stability challenges: - Removing implicit guarantees is politically and socially difficult due to SOE influence and potential layoffs. - Permitting more defaults could threaten financial stability in a system unaccustomed to risk. - Competitive pressures combined with implicit guarantees can encourage excessive risk taking by banks. - Recommendation: - Authorities should prioritize removal of implicit guarantees alongside broader interest-rate liberalization, sequencing reforms carefully to minimize financial-stability risks. ### Numerical solution algorithm (appendix highlights) - Asset grid discretized with 5,000 points. - Pareto entrepreneurial ability approximated with 10 grid points (Buera and Shin (2013) methodology). - Iterative procedure: 1. Guess k_soe / l_soe and lump sum tax T. 2. Compute r^{l}_{t,soe}, r^{l}_{t,poe}, w using (1), (2), and (15); solve value function and decision rules. 3. Simulate stationary distribution. 4. Update k_soe / l_soe until convergence. 5. Check government budget (10) and update T until balanced. - Note: with deposit ceiling fixed, r^d = ̄r. With liberalized deposits under monopolistic competition: r^d = r^{l}_{poe} / μ − χ. *Source: _wp15274 - 6.1 Liberalizing Deposit Interest Rates* ### 1. Introduction .................................................................................................... ### _wp15274 - 1. Introduction .................................................................................................... ### Overview and motivation - China’s financial sector reforms have lagged behind broader market-oriented reforms; two entrenched distortions are central: - administratively-controlled deposit interest rate ceilings; and - widespread implicit state guarantees for financial institutions and corporates (particularly SOEs). - These distortions have supported China’s growth by lowering the cost of capital and supporting an exceptionally high investment rate, but have produced mounting costs through misallocation of capital and the buildup of credit to low-return activities. - The paper calibrates a heterogeneous agent general equilibrium model à la Bewley-Aiyagari (aligned to Buera and Shin, 2013; and Quadrini, 2000) to Chinese data to simulate the effects of: - lifting deposit rate ceilings; and - removing implicit guarantees. ### Key quantitative facts from the background evidence - Deposit interest rate ceilings used to range from "0.35 percent" for demand deposits to "3.3 percent" for 1-year time deposits; banks were later allowed to offer rates up to "1.1 times" higher via a widening “float.” - Average effective deposit rate has been between "1-2 percent" and remained within this range even after full liberalization in October 2015. - Alternative saving products (WMPs and internet-based MMFs) offered yields in the range of "4 to 6 percent" and accounted for about "13 percent" of total bank deposits by mid-2015. - Bank cleanup costs in the early 2000s were estimated at more than "20 percent of GDP" by mid-2000s (Ma, 2006). - From the World Bank’s 2012 Enterprise Survey regression results (probit marginal effects): - POE: "0.123***" (column (1)) and other specifications show POE effects: "0.117***", "0.119***", "0.109***". - Manufacturing Sector: "0.0456***", "0.0667***", "0.0590***". - Retail: "0.134***", "0.0844**", "0.0760*". - log(employees): "-0.0390***", "-0.0371***". - log(years of operation): "-0.0144", "-0.0323*". - % of production exported: "-0.120***", "-0.0656*". - Growth of sales: "-0.0657**", "-0.0513*". - Observations: "2,734", "2,734", "2,514", "2,437". - Interpretation: Controlling for firm characteristics, a POE is "12 percent" more likely to be credit constrained; 23 percent of all firms in the survey were constrained. - Evidence of persistent borrowing-cost differentials: Ferri and Liu (2010) documented a spread of "200 bps" between SOE and POE borrowing rates in early to mid 2000s; more recent NBS data indicate a continued statistically significant spread in effective borrowing rates (illustrated by higher ratio of interest expenses to total liabilities for POEs vs SOEs). ### Mechanisms linking the two distortions - Deposit rate ceilings act as a financial repression (a tax on household savings) and reduce incentives for banks to improve efficiency. - Deposit ceilings and implicit guarantees coexisted and supported each other: implicit guarantees increase expected losses to the government and banks, while deposit rate ceilings provided a subsidy (below-market deposit funding) that helped rebuild bank balance sheets and share bailout costs with depositors. - The presence of implicit guarantees leads creditors to lend more and more cheaply to perceived-guaranteed entities (notably SOEs), reducing the price of risk and encouraging capital misallocation. ### Main simulation findings (model-based) - Removing deposit interest rate ceilings alone: - Would not by itself result in a more efficient allocation of credit. - Would reduce lending rates, increase capital intensity, and boost output primarily through higher capital stock rather than productivity gains. - With implicit guarantees still in place, both less efficient SOEs and more efficient POEs would expand given additional capital, leading to a reduction in total factor productivity (TFP) level. - Output gains could be smaller if other factors (including other Third Plenum reforms) lead to an offsetting reduction in savings. - Removing implicit guarantees: - Is the reform that leads to a more efficient allocation of capital and higher GDP, primarily by reducing the role of less efficient SOEs in the economy. - Policy implication emphasized: - Removing both distortions simultaneously is important because liberalizing deposit rates alone could reinforce distortions from implicit guarantees and could accentuate concerns about banking competition and risk-taking — a banking-stability channel not explicitly modeled here. ### Model structure (summary) - Time: discrete, annual periods. - Agents: four types — private agents (workers and entrepreneurs operating POEs), SOEs, banks, and a government. - POEs: face collateral constraints when seeking bank credit; hire labor and invest capital in a stochastic productivity process. - SOEs: enjoy better access to credit (lower interest rates and no collateral constraints). - Banks: state owned and monopolistically competitive. - Government: balanced budget; collects consumption taxes, lump-sum taxes, and profits from SOEs and banks; uses resources for public consumption and subsidies. *Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15274.pdf* ### 3.1 Households: Workers and Entrepreneurs ### 3.1 Households: Workers and Entrepreneurs ### Household environment and decision problem - Agents are heterogeneous in assets (a) and entrepreneurial ability (e). Assets and ability are the two state variables. - Entrepreneurial ability evolves with persistence: an agent keeps her ability with probability p_e ∈ (0, 1); otherwise she draws e′ from distribution μ(e′), invariant in time. - The endogenous distribution of agents at time t is Γ_t over (a,e). - Every period an agent chooses (i) whether to be a worker or run a firm (POE), and (ii) consumption and savings. - Recursive problem: - V_t(a_t,e_t) = max_{a_{t+1},c_t} U(c_t) + β E{V_{t+1}(a_{t+1},e_{t+1})} - s.t. a_{t+1} + (1+τ_c) c_t = max{w_t, π^{int}_t(a_t,e_t), π^{ext}_t(a_t,e_t)} + (1+r^d_t) a_t − T_t - a_{t+1} ≥ 0 - Definitions in the budget constraint: - V_t: household value function at time t - a_{t+1}: next-period asset holdings - c_t: consumption - τ_c: consumption tax - T_t: lump sum taxes - Income sources: - Deposits paying real interest r^d_t - Labor income w_t (inelastic labor supply; average working hours normalized to one) - Profits from POEs: π^{int}_t(a_t,e_t) if financed with internal funds; π^{ext}_t(a_t,e_t) if using external funds ### Occupational choice rule - Agent chooses to be a worker if w_t > max{π^{int}_t(a_t,e_t), π^{ext}_t(a_t,e_t)}; otherwise she becomes an entrepreneur. - If π^{int}_t(a_t,e_t) > max{w_t, π^{ext}_t(a_t,e_t)} → entrepreneur using only internal funds. - If π^{ext}_t(a_t,e_t) > max{w_t, π^{int}_t(a_t,e_t)} → entrepreneur using external funds. ### Firm profit: internal finance (POE using internal funds) - π^{int}_t(a_t,e_t) = max_{k^{int}_t,l^{int}_t} f(e_t,k^{int}_t,l^{int}_t) − (r^d_t + δ) k^{int}_t − w_t l^{int}_t - s.t. k^{int}_t ≤ a_t - Notation: - f(.) production function - δ depreciation rate - k^{int}_t capital invested - l^{int}_t labor demand ### Firm profit: external finance (POE using bank credit) - Banks require collateral; collateral is interest bearing. - Entrepreneur optimally deposits all assets (a) in the bank and gets credit paying lending rate r^{l}_{t,poe}. - π^{ext}_t(a_t,e_t) = max_{k^{ext}_t,l^{ext}_t} f(e_t,k^{ext}_t,l^{ext}_t) − (r^{l}_{t,poe} + δ) k^{ext}_t − w_t l^{ext}_t - s.t. a_t ≤ k^{ext}_t ≤ λ a_t - λ: collateral constraint parameter ### Production and entrepreneur ability distribution - Household/POE production: f(e,k,l) specified later (section 5). - Entrepreneurial ability μ(e) is invariant; later assumed Pareto with shape η and lower bound e_{low}: - μ(e) = 1 − (e / e_{low})^{−η} for all e ≥ e_{low} --- ### 3.2 SOEs - SOE sector: constant returns to scale Cobb-Douglas technology: - y_{t,soe} = A k_{t,soe}^{α} l_{t,soe}^{1−α} - SOEs access cheap credit: r^{l}_{t,soe} ≤ r^{l}_{t,poe}, and no collateral restrictions. - SOE problem: - max_{k_{t,soe}, l_{t,soe}} A k_{t,soe}^{α} l_{t,soe}^{1−α} − (r^{l}_{t,soe} + δ) k_{t,soe} − w_t l_{t,soe} - First order conditions (FOCs): - α (k_{t,soe} / l_{t,soe})^{α−1} = r^{l}_{t,soe} + δ (1) - (1−α) (k_{t,soe} / l_{t,soe})^{α} = w_t (2) - A: total factor productivity in SOE sector; α: capital-output elasticity --- ### 3.3 Banks - Continuum (measure unity) of monopolistic competitive banks, state owned; banks transfer profits to the government. - Each bank j supplies a different type of loan at interest r_t(j). - Aggregation of loans (CES with parameter μ > 1): - k^{ext}_{t,i} = [∫_0^1 k^{ext}_{t,i}(j)^{1/μ} dj]^{μ} - k_{t,soe} = [∫_0^1 k_{t,soe}(j)^{1/μ} dj]^{μ} - Bank-specific demand for credit: - k^{ext}_{t,i}(j) = ( r^{poe}_t(j) / r^{poe}_t )^{μ(1−μ)} k^{ext}_{t,i} (3) - k_{t,soe}(j) = ( r^{soe}_t(j) / r^{soe}_t )^{μ(1−μ)} k_{t,soe} (4) - Average lending rates (CES aggregator forms): - r^{poe}_t = [∫_0^1 r^{poe}_t(j)^{1/(1−μ)} dj]^{1−μ} - r^{soe}_t = [∫_0^1 r^{soe}_t(j)^{1/(1−μ)} dj]^{1−μ} - Resource constraint for bank j (deposit scarcity): - k_{t,soe}(j) + ∫_{P_t,ext} k^{ext}_{t,s}(j) ds ≤ D_{t,j} (5) - D_{t,j}: deposits available to bank j at time t - P_{t,ext}: measure of producers demanding external funds - Bank profit when lending to POE i: - π^b_{t,i}(j) = max_{r^{i}_t(j)} { r^{i}_t(j) k^{ext}_{t,i}(j) − (r^d_t + χ) k^{ext}_{t,i}(j) } s.t. (3) and (5) - χ: intermediation costs; r^d_t: deposit interest rate - First-order implication: - r^{i}_t(j) = μ (r^d_t + χ) + ξ_{t,j} (6) - ξ_{t,j} : non-negative multiplier from resource constraint (5); ξ_{t,j} > 0 only when deposit interest rate ceiling binds - μ represents the markup banks would charge once deposit rate is liberalized - Bank profit when lending to SOEs: - π^b_{t,soe}(j) = max_{r^{soe}_t(j)} { r^{soe}_t(j) k_{t,soe}(j) − (r^d_t + χ − κ) k_{t,soe}(j) } s.t. (4) and (5) - κ: subsidy per unit of capital lent to SOEs (reduced-form implicit guarantee) - Implication: - r^{soe}_t(j) = μ (r^d_t + χ − κ) + ξ_{t,j} (7) - Symmetric banking equilibrium (drop j): - r^{poe}_t = μ (r^d_t + χ) + ξ_t (8) - r^{soe}_t = μ (r^d_t + χ − κ) + ξ_t (9) --- ### 3.4 Government - Government balanced budget in stationary equilibria: - G_t + κ ∫_0^1 k_{t,soe}(j) dj = T_{c,t} + T_t + π^b_t (10) - π^b_t = ∫_0^1 π^b_{t,soe}(j) dj + ∫_{P_{t,ext}} ∫_0^1 π^b_{t,i}(j) dj di - Left-hand side: government consumption G_t and subsidies κ times amount lent to SOEs - Right-hand side: consumption taxes T_{c,t}, lump-sum taxes T_t, and bank profits π^b_t --- ### 3.5 Distortions - Two key distortions highlighted: 1. Deposit interest rate ceiling: deposit rate in China cannot exceed a government-set r̄. Empirically, average deposit interest rate is very low and appears binding; model equilibrium with ceiling sets r^d_t = r̄. - With ceiling binding (ξ_t > 0), lending rates: - r^{poe}_t = μ ( r̄ + χ ) + ξ_t (11) - r^{soe}_t = μ ( r̄ + χ − κ ) + ξ_t (12) - When r^d_t is liberalized, multiplier ξ_t = 0 and lending rates become: - r^{poe}_t = μ ( r^d_t + χ ) (13) - r^{soe}_t = μ ( r^d_t + χ − κ ) (14) - Ceiling generates an additional premium ξ_t in observed lending spreads. 2. Implicit guarantees to SOEs: modeled as subsidy κ per unit lent to SOEs; reduces lending rate to SOEs relative to POEs. - Equilibrium condition equating marginal profit implies: - r^{soe}_t + μ κ = r^{poe}_t (15) - Given μ, κ is pinned down by the SOE vs. POE spread. --- ### 4 Equilibrium (definitions and market clearing) - Two equilibrium types: - With binding deposit interest ceiling: deposit rate fixed to r̄; lending rates clear capital market via (11) and (12); scarcity reflected in ξ_t. - Without ceiling: both deposit and lending rates clear market via (13) and (14); ξ_t = 0. - State spaces: A := [0,∞) (assets), E (support of μ(e)). - Definition 1: Competitive equilibrium with binding deposit interest ceiling r̄ is a set of policy rules {a_{t+1}(a_t,e_t), c_t(a_t,e_t), l^{int}_t(a_t,e_t), l^{ext}_t(a_t,e_t), k^{int}_t(a_t,e_t), k^{ext}_t(a_t,e_t), k_{t,soe}, l_{t,soe}}, prices {r^{l}_{t,poe}, r^{l}_{t,soe}, w_t}, multipliers {ξ_t} and fiscal policy {τ_c, G_t, T_t, κ} such that given Γ_0 and for all t: 1. Given r^{l}_{t,poe}, w_t, policies l^{int}_t(...), l^{ext}_t(...), k^{int}_t(...), k^{ext}_t(...) solve POE problem. 2. Given r^{l}_{t,soe}, w_t, policies k_{t,soe}, l_{t,soe} satisfy (1) and (2). 3. {a_{t+1}(...), c_t(...)} solve household problem. 4. r^{l}_{t,poe}, r^{l}_{t,soe}, ξ_t satisfy (11) and (12). 5. Markets clear: - k_{t,soe} + ∫∫_P [k^{int}_t(a_t,e_t) + k^{ext}_t(a_t,e_t)] Γ_t(da,de) = ∫∫ a_t Γ_t(da,de) - l_{t,soe} + ∫∫_P [l^{int}_t(a_t,e_t) + l^{ext}_t(a_t,e_t)] Γ_t(da,de) = ∫∫_P c Γ_t(da,de) - P := {(a_t,e_t) ∈ A × E : w_t ≤ max{π^{int}_t(a_t,e_t), π^{ext}_t(a_t,e_t)}} 6. Government budget constraint (10) holds. 7. Distribution transition: - Γ_{t+1}(A×E) = ∫_{A×E} Q((a,e),A×E) Γ_t(da,de) - Q((a,e),A×E) = I{a_{t+1}(a,e) ∈ A} [ p_e I{e ∈ E} + (1−p_e)(1−I{e ∈ E}) ∫_E μ(de) ] - Definition 2: Competitive equilibrium without deposit interest ceiling: same conditions as Definition 1 except condition 4 replaced by r^{l}_{t,poe}, r^{l}_{t,soe} and r^d_t satisfy (13) and (14). - Stationary equilibria: Γ_{t+1} = Γ_t for all t. - Definition 3: Stationary equilibrium: competitive equilibrium such that Γ^* invariant and satisfies Γ^*(A×E) = ∫_{A×E} Q((a,e),A×E) Γ^*(da,de) --- ### 5 Functional forms and calibration - Utility and production functional forms (following Buera and Shin (2013)): - U(c) = c^{1−σ} / (1−σ) - f(e,k,l) = e (k^{α} l^{1−α})^{1−ν} - Parameters introduced: - σ: relative risk aversion parameter - α: capital share (assumed equal to SOE sector) - ν: Lucas (1978) span-of-control parameter - Entrepreneur ability μ(e) drawn from Pareto with shape η and lower bound e_{low} - μ(e) = 1 − (e / e_{low})^{−η} for e ≥ e_{low} - Calibrated parameter choices cited as standard: - σ = 1.5 (implying EIS = 0.6) - δ = 0.06 - α = 0.4 - ν = 0.21 - p_e = 0.894 - A = 1 (normalization for SOE TFP) - Banking sector markup μ difficult to identify under ceiling; literature suggests Lerner indexes between 5 percent and 15 percent for BRIC countries; baseline μ = 1.1 chosen (median). ### Key calibrated parameters and matching moments (Table 2 highlights) - Moments matched and calibrated parameter values (as presented): - Nominal deposit interest rate: 330 bps — Model: 330 bps — ̄r = 0.008 - SOEs vs POEs spread: 200 bps — Model: 200 bps — κ = 0.02/1.1 - POEs nominal loan rate: 786 bps — Model: 793 bps — β = 0.892 - Collateral (% of loan): 137% — Model: 137% — λ = 1.73 - Banks overhead cost (% of assets): 1.15% — Model: 1.15% — χ = 0.0115 - VAT rate: 17% — Model: 17% — τ_c = 0.17 - Government consumption (% GDP): 13.5% — Model: 13.5% — g = 0.135 - Labor share SOEs: Data 18.4% — Model 21.5% - Labor held by top 5% private firms: Data 46.0% — Model 46.5% — e_{low} = 0.945 - Labor held by top 10% private firms: Data 57.2% — Model 58.8% — η = 3.63 - Labor held by top 20% private firms: Data 70.4% — Model 70.7% - Labor held by top 30% private firms: Data 78.9% — Model 77.6% - Labor held by top 40% private firms: Data 84.9% — Model 82.2% - Additional calibration notes: - Nominal deposit interest rate set at ceiling as of summer 2014. - 200 bps SOE-POE spread taken from Ferri and Liu (2010). - POE lending rate computed using CEIC Data and NBS. - Collateral percent from World Bank Enterprise Survey for China (2012). - Banks’ overhead cost (% assets) from Beck and others (2000), average between 2000 and 2011. - Annual inflation rate assumption: 2.5% - VAT and government consumption ratios from stated data sources; labor distribution from NBS for 2009. ### 6 Results (model application) - The calibrated model will be used to assess the impact of: - (i) liberalizing the deposit interest rate - (ii) removing implicit guarantees (κ) *Source: _wp15274 - 3.1 Households: Workers and Entrepreneurs* ### 6.1 Liberalizing Deposit Interest Rates ### 6.1 Liberalizing Deposit Interest Rates ### Impact of liberalizing the deposit rate ceiling - Comparative stationary equilibria: with deposit rate ceiling versus without. - Without the ceiling: - Nominal deposit interest rate rises from the ceiling of 330 bps to around 580 bps. - Lending interest rate expected to decline by around 50 bps for both POEs and SOEs. - Capital intensity in the economy increases. - Capital stock boost raises GDP by around 4 percent. - Total TFP falls (reduction in TFP), driven mainly by: - Diminishing marginal returns to capital (higher capital stock → lower productivity). - Additional capital allocated to both productive POEs and less productive SOEs that enjoy implicit guarantees, lowering overall TFP. - Banking margins: - Movement in deposit and lending rates shrinks banking margins. - Under monopolistic competition, profits (margins) are determined by the markup level μ; the new equilibrium depends on μ (degree of competition). - Risks not captured by the model: - Model does not account for possible risk-shifting behavior in banks; risk shifting could change portfolio composition and tend to increase lending interest rates. - Stronger competition can increase incentives for banks to take risk, harming financial stability (referenced literature: Vives, 2010; country examples: US in the 1980s, Spain). ### Role of banking competition (markup μ) - Equilibrium outcomes vary with markup μ: - When μ = 1.3 (lower competition / higher markup): deposit interest rate increases only to a level of 500 bps. - When μ = 1.1 (higher competition / lower markup): deposit rate goes up to around 580 bps. - Lending interest rates remain higher when competition is lower. - Effects on GDP, TFP and capital-output ratio are reduced when banking competition is lower. - Summary: Greater competition amplifies the increase in deposit rates and the decrease in lending rates and magnifies macroeconomic effects; low competition mutes these effects. ### Removing implicit guarantees in addition to liberalizing deposit rates (section 6.2) - Distinct effects: - Deposit rate liberalization: quantity-based gains (output grows because of higher capital stock). - Removing implicit guarantees: efficiency-based gains (better allocation of capital and labor). - With both reforms (compare second and third bars in Figure 9): - Deposit rate is pushed downward (relative to no implicit-guarantee case). - POEs experience a significant fall in their lending rate. - Capital intensity decreases (less capital available). - Labor share of SOEs drops from 22 percent to around 12 percent. - GDP increases by around 1 percent. - Total TFP increases by 3 percent. - Intuition (Figure 10): - Removing implicit guarantees shifts demand for capital left (SOEs lose subsidy), reducing deposit and POE lending rates. - Lower POE lending rates allow more POEs access to credit → increases income and savings → shifts savings curve right. - In the calibration, net effect is reduced capital intensity (demand-side reduction dominates). - SOE lending rate effect: - SOEs witness a slight rise in their lending interest rates (around 20 bps). - This increase is less than 200 bps (the subsidy rate removed) due to general equilibrium effects that lower r_l_poe and thus partially offset the rise in r_l_soe. - From equation (15) in the model: r_l_soe = r_l_poe − μκ, where κ is the subsidy rate and r_l_poe is the POE lending rate. ### Interaction with other reforms and savings elasticity (section 6.3) - Reforms such as strengthening social safety nets and lifting the one-child policy could shift aggregate savings curve to the left. - The paper models these via a shock to the discount factor β. - Illustrative calibration: shock calibrated so total stock of savings does not change after interest rate liberalization (Figure 11). - Key intuition: - Economy’s reaction depends on elasticity of savings to interest rate changes: more inelastic savings → lower impact of interest rate liberalization. - If savings do not change, stock of capital available for POEs and SOEs remains fixed; GDP impact comes from better allocation of resources. - With fixed savings, removing implicit guarantees still increases total TFP even if capital stock falls relative to pre-reform levels. - Comparative outcomes: - When capital resources are fixed, the impact on GDP from liberalization is reduced relative to the case where savings expand. - Adding removal of implicit guarantees with fixed savings yields higher increases in total TFP than the case without fixed savings. ### Policy implications and risks (Conclusion summary) - Progress to date: - Interest rate liberalization largely completed, though PBC retains some control via “window guidance”. - Removal of implicit state guarantees has been slower. - Key findings: - Partial reforms (liberalized deposit rates while implicit guarantees remain) are insufficient and may worsen inefficiency by channeling larger savings into less efficient SOEs and entities with public support. - Removing implicit guarantees is key to more efficient growth: it would reduce savings and capital intensity while reallocating credit to more efficient private enterprises and boosting total factor productivity. - Risks and trade-offs: - Removing implicit guarantees is politically and socially difficult due to SOE lobby and concerns about layoffs and social stability. - Permitting more defaults could have negative repercussions for financial stability in a system unaccustomed to risk. - Competitive pressures combined with implicit guarantees could lead to excessive risk taking and threaten financial stability. - Recommendation: - Authorities should prioritize removal of implicit guarantees alongside broader liberalization, treading carefully to minimize threats to financial stability — the time to act is now. ### Appendix: Algorithm to compute stationary equilibria - Numerical setup: - Discretize asset space using a grid of 5,000 points. - Pareto distribution of entrepreneurial ability approximated using methodology from Buera and Shin (2013) with 10 grid points. - Iterative procedure (following Quadrini (2000)): 1. Guess an initial k_soe / l_soe ratio and a lump sum tax T. 2. Using equations (1), (2), and (15) compute r_l_soe, r_l_poe and w. Solve the value function and obtain decision rules. 3. Using policies, get the stationary distribution of agents by simulation. 4. With the stationary distribution and market clearing conditions, get a new ratio k*_soe / l*_soe. If difference exceeds tolerance, update guess and repeat from step 1. 5. Using the stationary distribution, check budget constraint (10). If balance ≠ zero, update lump sum tax and repeat until convergence. - Note on deposit rate when ceiling is fixed: - For equilibria with deposit interest rate fixed at the ceiling, r_d is fixed at ̄r. - Note on liberalized deposit rate under monopolistic competition: - r_d = r_l_poe / μ − χ, where χ is intermediation cost for banks and μ is the markup. *Italic: Source — _wp15274 - 6.1 Liberalizing Deposit Interest Rates* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15274.pdf_