Retirement in the Shadow (Banking) 1 Guillermo Ordoñez 1 University Facundo Piguillem2 of Pennsylvania 2 EIEF March 19, 2015 Motivation I Does shadow banking serve any “real” function for the economy? I Did it generate the lending boom in the 90’s and early 2000? This paper I I Incorporates: I Costly intermediation: traditional and shadow banking I Heterogeneous bequest motives: I Uncertain life spans: saving for retirement Main mechanism: in equilibrium households I with low bequest motive want insurance (e.g., buy annuities) I with high bequest motive want to self insure (e.g., buy equities) This paper I I I Incorporates: I Costly intermediation: traditional and shadow banking I Heterogeneous bequest motives: I Uncertain life spans: saving for retirement Main mechanism: in equilibrium households I with low bequest motive want insurance (e.g., buy annuities) I with high bequest motive want to self insure (e.g., buy equities) Why? I Care differently about accidental bequests This paper I I I Incorporates: I Costly intermediation: traditional and shadow banking I Heterogeneous bequest motives: I Uncertain life spans: saving for retirement Main mechanism: in equilibrium households I with low bequest motive want insurance (e.g., buy annuities) I with high bequest motive want to self insure (e.g., buy equities) Implies: High bequest motive hh’want to borrow from low I If intermediation between households is costly I ⇒ return differential on assets Large increase in financial intermediation: 1980-2005 Private debt /GNP: Table D.3 Flow of funds. Excluding Government debt 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 1961 1959 1957 1955 1953 1951 1949 1947 1945 0.4 go Large increase in retirement funds Life expectancy increased from 74 to 79 years Household's assets held in retiretment funds Table L100 flow of funds Line 20 12 1.2 Pensions assets over GDP 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 No improved efficiency in the financial sector Value added per unit of loan. Philippon (AER forthcoming) 3.0% Level based Quality Adjusted 2.5% 2.0% 1.5% 1.0% 0.5% 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 1961 1959 1957 1955 1953 1951 1949 1947 0.0% 1945 I Big drop in Liquidity cost Use NIPA, Flow of Funds and Philippon data to construct: "Liquidity cost" of Financial Sector 2.5% Final-Level based Final-Quality adjusted 2.0% 1.5% 1.0% 0.5% 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 1978 1976 1974 1972 1970 1968 1966 1964 1962 1960 1958 1956 1954 1952 1950 1948 0.0% 1946 I Main Findings I I I Spread between borrowing and lending fell 1 percentage point or, the intermediation cost fell by 33% I Borrowing and lending increased from 1GDP to 1.7GDP I Total assets explode from 4GDP to 10GDP We account for these facts I Almost all the reduction in spread due to Liquidity I No increase in efficiency. I Increase in life expectancy key to get the right direction of prices Road Map 1. Accounting for the change in intermediation cost. 2. Model economy 2.1 Individual Decisions 2.2 Intermediation: traditional and shadow banking 2.3 Equilibrium 3. Solution and Parameters 4. Results 5. Conclusions Accounting for spreads I A fundamental equation Gross output = Value added + Cost of inputs Int. received − Bad debt = VA + Int. paid + other services fre + (1 − f )rL − sb (1 + re ) = φ̂ + (rd + rs ) I Where: I I I I I I I I re : interest rate charged on a loan. rL : return on liquid assets. 1 − f : proportion of assets invested in liquid assets. sb : proportion of assets that are lost. φ̂: value added per unit of loan. rd : return to investor (or depositor) rs : value of other services to investor Now we define r = rd + rs : opportunity cost for investor. Accounting for spreads I Reorganize equation fre + (1 − f )rL − sb (1 + re ) = φ̂ + (rd + rs ) re − r | {z } Interest spread I = φ̂ |{z} value added + sb (1 + re ) + (1 − f )(re − rL ) {z } | {z } | risk component liquidity component We focus on last the component: (1 − f )(re − rL ) I If liquid asset have same return as loans re = rL , no liquidity cost. I If lenders do not need liquidity, f = 1, no liquidity cost. I Cost could depend on level. e.g, if rL = 0 (cash), spread increases with the level of interest rates. Accounting for spreads: first look I re=interest received/private debt. Table 7.11 Line 28/Table D.3. I r =interest paid deposits/deposits. Table 7.11 Line 5/Table L109. 16% r=Average return on deposits 14% re = Average return on loans re-r, Naïve spread 12% 10% 8% 6% 4% 2% 2013 2011 2009 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 1961 1959 0% Accounting for spreads: second look I Financial sector has changed. Less traditional banking, more non standard banking. Around 1/3 of “banks” liabilities are deposits. Deposits over private debt. Flow of funds: Table L109/Table D.3 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 I 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 1961 1959 1957 1955 1953 1951 1949 1947 1945 0.0 Plus, financial institutions offer several services besides rd . We use “Service Furnished without payment” (Table 2.4.5 line 88) Accounting for spreads: second look r =(interest paid+Service Furnished without pay.)/private debt. (Use Table 7.11 Line 4 instead of line 5) 16% r=Average return on liabilities 14% re = Average return on loans re‐r, Less naive spread 12% 10% 8% 6% 4% 2% 2013 2011 2009 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 1961 0% 1959 I Accounting for spreads: risk component I Table 7.1.6 Line 12 provides “Bad debt expenses” for corporate business. I Not all corporate business are Financial. We follow Mehra et al (2011) assigning half to Financial. I Assigning all of it does not change anything but the level. 3.0% 2.0 Bad deb expenses B a 2.5% d 1.8 Private intermediated quantities 1.6 D 1.4 e b t 12 1.2 d e 2.0% b t v e r G D P o 1.0 v e 0.8 r 1.5% 1.0% 0.6 0.4 ( 0.5% & ) 0.2 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 0.0 1945 1947 1949 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 0.0% G D P Accounting for spreads: what is left After taking risk component out we are left with: r −r |e {z } = Risk adjusted spread φ̂ |{z} value added + (1 − f )(re − rL ) {z } | liquidity component 5% risk adjusted spread j p 4% 4% 3% 3% 2% 2% 1% 1% 2011 1 2009 9 2007 7 2005 5 2003 3 2001 1 1999 9 1997 7 1995 5 1993 3 1991 1 1989 9 1987 7 1985 5 1983 3 1981 1 1979 9 1977 7 1975 5 1973 3 1971 1 1969 9 1967 7 1965 5 1963 3 1961 1 0% 1959 9 I Accounting for spreads: what is left I Why did the spread fall? Was efficiency, φ̂, or liquidity (1 − f )(re − rL )? I Did the IT revolution reduce the spread or was it liquidity cost? Accounting for spreads: what is left I Why did the spread fall? Was efficiency, φ̂, or liquidity (1 − f )(re − rL )? I Did the IT revolution reduce the spread or was it liquidity cost? Philippon (2015) shows φ̂ has been constant for more than 100 years. I He also shows that financial intermediation technology exhibits CRS. I Accounting for spreads: what is left I Why did the spread fall? Was efficiency, φ̂, or liquidity (1 − f )(re − rL )? I Did the IT revolution reduce the spread or was it liquidity cost? Philippon (2015) shows φ̂ has been constant for more than 100 years. I He also shows that financial intermediation technology exhibits CRS. I 3.0% Level based Quality Adjusted 2.5% 2.0% 1.5% 1.0% 0.5% 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 1961 1959 1957 1955 1953 1951 1949 1947 1945 0.0% Accounting for spreads: the matter in issue What is left after subtracting value added? Liquidity component 2.5% Final Level based Final‐Level based Final‐Quality adjusted 2.0% 1 5% 1.5% 1.0% 0.5% 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 1978 1976 1974 1972 1970 1968 1966 1964 1962 1960 1958 1956 1954 1952 1950 1948 0.0% 1946 I Accounting for spreads: some numbers I From 1980 to 2007 the spread fell around 1%, all of it due to a reduction in the liquidity component. I In 2007 the liquidity cost was almost 0%! I Assume cost in traditional baking, φTB , equal to cost in shadow banking φSB . Then, φTB = φSB = φ̂ (Philippon claims this is the case.) I Shadow banking has no liquidity provisions, fSB = 1, then (share TB) × (1 − f )(re − rL ) + (share SB) × 0 = Liquidity cost I Share TB fell from 2/3 to 1/3 from 1980 to 2007 (Figure with dep.) I ABS used as liquid assets reducing re − rL in TB (numbers?) Liquid assets in depository institutions Flow of Funds Table L109. Liquid assets of depository institutions over total deposits. 0.3 Cash plus deposits in FED P r o p o r t i o n Cash+plus deposits in Fed plus Treasuries p p p 0.25 Liquidity: including ABS 0.2 0.15 o f 0.1 d e p o s i t s 0.05 0 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 I Summary I Big increase in borrowing and lending. I Large drop in financial intermediation spread. I All the fall due to “liquidity cost” I SB have direct impact: share increased from 1/3 to 2/3 I and also indirect impact. It generated liquid assets with larger rL I Who or what and why provided the additional funds? After any new borrower there has to be a lender! I How important is this change? Isn’t 1% too little? Summary I Big increase in borrowing and lending. I Large drop in financial intermediation spread. I All the fall due to “liquidity cost” I SB have direct impact: share increased from 1/3 to 2/3 I and also indirect impact. It generated liquid assets with larger rL I Who or what and why provided the additional funds? After any new borrower there has to be a lender! I How important is this change? Isn’t 1% too little? Answer: 1% is huge. Population is aging, need for retirement insurance. Answering the question I We build on Mehra et al. (2011) I OLG with financial intermediaries. I We show that the decrease in intermediation cost plus the increase in life expectancy generates the observed changes in aggregate quantities and prices. I Reduction in spread is welfare improving. Help with provision of insurance. I Disclaimer: we are abstracting from potential instability of shadow banking. There are enough papers about it. Model: Technology and Population I Overlapping generations with Neoclassical technology Yt = Ktθ (zt Lt )1−θ zt+1 = (1 + γ)zt I I θ: share of capital in output γ: growth rate of labor productivity Model: Technology and Population I Overlapping generations with Neoclassical technology Yt = Ktθ (zt Lt )1−θ zt+1 = (1 + γ)zt I I I θ: share of capital in output γ: growth rate of labor productivity Measure 1 of people are born every period I I Each one with a different intensity of bequest motive α ≥ 0 α ∼ f (α) Agents I Agents order consumption according to U(α, c, b) = T X j=0 β j log cj + ∞ X β j (1−δ)j−T −1 [(1−δ) log cj +δα log bj ] j=T +1 I α ≥ 0: strength of bequest motive I 0 < δ < 1: probability of death I Agents work for T years, earn wage ωj and then retire Agents I Agents order consumption according to U(α, c, b) = T X j=0 β j log cj + ∞ X β j (1−δ)j−T −1 [(1−δ) log cj +δα log bj ] j=T +1 I α ≥ 0: strength of bequest motive I 0 < δ < 1: probability of death I Agents work for I Bequest bj , equally distribute amongst agents of age I At age T years, earn wage ωj and then retire j = TI < T receive inheritance b̄ j = TI < T Investment Decision I Assumption: Restricted choice for households I Agents choose whether to sign or not annuity contract at age j =0 Investment Decision I Assumption: Restricted choice for households I Agents choose whether to sign or not annuity contract at age j =0 Then agents have two options: I Strategy A ⇒ buy annuities I I I Benefit: Insurance against risk of living long time Cost: Low return on assets (r ) Strategy B ⇒ go on your own I I Benefit: High return on assets (re ) Cost: No insurance Investment Decision I Assumption: Restricted choice for households I Agents choose whether to sign or not annuity contract at age j =0 An annuity contract between an agent and the intermediary is: I While working: payment made by agent to intermediary I When retired: payment made by intermediary to agent I I cj if alive bj when she dies Investment Decision I Assumption: Restricted choice for households I Agents choose whether to sign or not annuity contract at age j =0 Let, interest differential: φ = re − r Propositions 1 and 2 If φ ≤ φ ≤ φ̄, there exists α∗ > 0 such that, I if α < α∗ follow strategy A I if α ≥ α∗ follow strategy B I From now on assume: µa 1 − µa F (α) = 0 if α = 0 if α = α̂ > 0 otherwise Agent Problem: Strategy A (annuity) max T X β t log cj + j=0 ∞ X β j (1 − δ)j−T −1 [(1 − δ) log cj + δα log bj ] j=T +1 subject to T X j=0 ∞ X ct (1 − δ)j−T −1 [(1 − δ)cj + δbj ] + ≤ v0A j (1 + r ) (1 + r )j j=T +1 v0A = T −1 X j=0 I r : household lending rate I v0A : “wealth” at time 0 I ωj : wage at age j b̄ (1 − τ )ωj + j (1 + r ) (1 + r )TI Solution annuity strategy: consumption (α low) 0.045 Working age 0.040 Retirement age Type A 0.035 Consumption 0.030 0.025 0.020 0.015 0.010 0.005 0.000 22 26 30 34 38 42 46 50 54 Age 58 62 66 70 74 78 82 Solution annuity strategy: assets accumulation (α low) 20.0 Net Worth/Annual wage 16.0 12.0 Type A 8.0 4.0 0.0 22 26 30 34 38 42 46 50 54 Age 58 62 66 70 74 78 82 Agent Problem: Strategy B (no-annuity) I Two problems: before retirement and after that Agent Problem: Strategy B (no-annuity) I Two problems: before retirement and after that I After retirement V (w ) = max{log c + (1 − δ)βV (w 0 ) + δβα log w 0 } subject to c+ I w0 ≤w (1 + re ) re : return on equity and the household borrowing rate Agent Problem: Strategy B (no-annuity) I Solution after retirement V (w ) = ν̄1 (α) + ν̄2 (α) log w I where ν̄2 (α) = 1 + αβδ 1 − (1 − δ)β Agent Problem: Strategy B (no-annuity) I Solution after retirement V (w ) = ν̄1 (α) + ν̄2 (α) log w I where ν̄2 (α) = I 1 + αβδ 1 − (1 − δ)β The optimal consumption and implicit bequest strategies are: c = w /ν̄2 (α) w 0 = (1 + re )(w − c) I Bequests are: bj = wj j ≥T Agent Problem: Strategy B (no-annuity) I At entry in labor force max T −1 X β j logcj + β T V (wT ) j=0 subject to T −1 X j=0 v0B cj wT + ≤ v0B (1 + re )j (1 + re )T = T −1 X j=0 (1 − τ )ωj b̄ + (1 + re )j (1 + re )TI Solution No-annuity strategy: consumption (α high) 0.045 Working age Retirement age 0.040 Type B 0.035 Consumption 0.030 0.025 0.020 0.015 0.010 0.005 0.000 22 26 30 34 38 42 46 50 54 Age 58 62 66 70 74 78 82 Solution No-annuity strategy: assets accumulation (α high) 20.0 Type B Net Worth/Annual wage 16.0 12.0 8.0 4.0 0.0 22 26 30 34 38 42 46 50 54 Age 58 62 66 70 74 78 82 Lifetime pattern of consumption Consumption 0.045 0.040 Type A 0.035 Type B Working age Retirement age 0.030 0.025 0.020 0.015 0.010 0.005 0.000 22 26 30 34 38 42 46 50 54 Age 58 62 66 70 74 78 82 Cross sectional distribution of consumption 0.025 Type B Retirement age Type A 0.020 consumption Working age 0.015 0.010 0.005 0.000 22 26 30 34 38 42 46 50 54 age 58 62 66 70 74 78 82 Intermediation: traditional banking B A I Intermediary pays r for the amount borrowed I Lend to other hh’s, with proportional cost of φ, or to the gov’t at zero cost Intermediation: traditional banking B A I Intermediary pays r for the amount borrowed I Lend to other hh’s, with proportional cost of φ, or to the gov’t at zero cost Assets Liabilities Government Debt (D G ) Assets of type A (W A ) Private Debt (D B ) Intermediated services D B (re − r ) Net Worth = 0 Assets = D B (1 + re ) + D G (1 + r ) W A = D B (1 + r ) + D G (1 + r ) Closing the economy I Fiscal policy to keep DtG fixed and tax τ on labor income I G − DtG ) τ ωt Lt = rDtG − (Dt+1 D G : government debt I Competitive goods market (δk : depreciation rate) δk + re = FK (Kt , zt Lt ) ωt = FL (Kt , zt Lt ) I Competitive intermediaries I re = φ + r Aggregates I At period t, measure of agents age j of type i = A, B is given by µi if j ≤ T µit,j = (1 − δ)j−T −1 µi if j > T I Use measures to get aggregates, i = A, B Ct (α̂, τ, b̄) = ∞ XX i Wti (α̂, τ, b̄) = Bt (α̂, τ, b̄) = j=1 ∞ XX i = T wt,j (αi ; τ, b̄)µit,j j=1 ∞ X j=T +1 Lt ct,j (αi ; τ, b̄)µit,j δbt,j (α̂; τ, b̄)µBt−1,j−1 (stationary) Equilibrium: An equilibrium is {c i , w i , b i , K , α∗ }, prices re , r and fiscal policy such that, 1. Households maximize I α̂ ≥ α∗ 2. Intermediary maximize 3. Government budget constraint holds 4. Market clearing Market Clearing: I A B Feasibility Y = C (α̂, τ, b̄) + X + φ(K − I W B (α̂, τ, b̄) ) 1 + re Assets market W A (α̂, τ, b̄) W B (α̂, τ, b̄) + = DG + K 1+r 1 + re I Bequest=inheritance b̄ = (1 + γ)TI B(α̂, τ, b̄) Solution Strategy (THIS MUST BE CHANGED) I Balanced Growth analysis I All variables except interest rates and L grow at a rate γ Solution Strategy (THIS MUST BE CHANGED) I Balanced Growth analysis I All variables except interest rates and L grow at a rate γ I Given τ , b̄ and α̂ compute individual decisions I In balanced Growth: ct,j (α; τ, b̄) = ct+k,j+k (α;τ,b̄) (1+γ)k Solution Strategy (THIS MUST BE CHANGED) I Balanced Growth analysis I All variables except interest rates and L grow at a rate γ I Given τ , b̄ and α̂ compute individual decisions I In balanced Growth: ct,j (α; τ, b̄) = I Use above to compute aggregates I Solve for α̂, τ and b̄ such that feasibility, assets markets and indifference are satisfied ct+k,j+k (α;τ,b̄) (1+γ)k Calibration I Calibrate model economy to 1980. I φ = 0.03 (Spread from NIPA discussed before) Calibration I Calibrate model economy to 1980. I φ = 0.03 (Spread from NIPA discussed before) Parameters associated with individuals I β = 0.99 I δ = 0.08 ⇒ post retirement life expectancy of 12.5 years Calibration I Calibrate model economy to 1980. I φ = 0.03 (Spread from NIPA discussed before) Parameters associated with individuals I β = 0.99 I δ = 0.08 ⇒ post retirement life expectancy of 12.5 years Agents enter the workforce at 22 I TI = 30 ⇒ inheritance is received at 52 I T = 40 ⇒ retirement at 62 Calibration Goods production parameters I θ = 0.3. Share of capital in output I γ = 0.02. Consistent with observations in labor productivity I δk = 0.04. Consistent with capital output ratio = 3. I D G /Y = 0.62. From data (with population growth is smaller) I µA = 0.76. Measure of annuity guys I α̂ = 3. These last two determined equilibrium bequest. Calibration implies I r = 0.03 and re = 0.06 Summary of Aggregate Results ∗ Economy Indirect Equity holding∗ Direct Equity holding∗ Model 0.75 0.25 Data 0.72 0.28 National Accounts Output Consumption Intermediated Services 1 0.79 0.03 1 071 0.027 Net Worth Total Government Debt/GDP 3.62 0.61 4.59 0.53 Bequest/GDP Households Debt/GDP 0.045 1.00 0.027 0.99 Equities directly hold/total financial assets. Table B.100e of flow of funds Discussion Fig. 1 I With populations growth all the numbers match almost exactly I There are some implicit government liabilities I Density of types and α̂ jointly determine bequest Counterfactual: the 2005 economy Economy Interm. Cost Life expectancy r re Benchmark (1980) 3% 0.08 New δ 3% 0.06 New φ 2% 0.08 2005 2% 0.06 0.030 0.060 0.023 0.053 0.032 0.052 0.026 0.046 National Accounts Capital stock Output Net Worth Type A (deposits+pensions) Type B (equity) Total financial Assets (WA + D G ) Data Fin. Assets (Table L100) 3.00 1.00 3.23 1.03 3.26 1.04 3.47 1.07 1.62 2.00 2.24 2.40 1.72 2.11 2.34 2.27 1.61 2.9 2.36 1.73 3.00 3.30 Government Debt/Y Bequest/Y Households Debt/GDP Data on debt 0.61 0.045 1.00 1.00 0.62 0.045 1.11 0.62 0.035 1.65 0.62 0.036 1.74 1.73 Discussion I What do we get from the table? I Drop in φ gives a large increase in debt, not enough, but wrong movement in r I Increase in life expectancy give little additional debt, but right movement in r I Both together deliver the right debt and movement in r I Total household financial assets a bit low. The change is the right magnitude. Adding land in K will make all numbers right. I Population growth rate plus introducing right government debt would make all numbers be almost perfect. Type A Intermed. back B Assets Liabilities Pension Assets 0 Net Worth Pension Assets = D G (1 + r ) + D B (1 + r ) Net Worth = Pension Assets Type B Intermed. back A Assets Liabilities K (1 + re ) D B (1 + re ) Net Worth Net Worth = W0B (τ, b̄) = (K − D B )(1 + re ) Quantities intermediated = D B Lorenz curve for consumption, total net worth and capital back 100% 90% Percent of C, K and NW 80% 70% 60% 45° Line 50% 40% Consumption 30% Net Worth 20% Capital 10% 0% 0% 10% 20% 30% 40% 50% 60% Percent of Population 70% 80% 90% 100% Balance Sheet of Borrowers Assets Corporate Capital Non Corporate Capital Real Estate Liabilities Corporate Debt Non Corporate Debt Mortgages Net Worth What could go wrong? Difference of utilities at α = 0 Private Debt over GDP back