Quantifying the Impact of Financial Development on Economic Development

Quantifying the Impact of Financial Development on Economic Development Jeremy Greenwood, Juan M. Sanchez, Cheng Wang (RED 2013) Presented by Beatriz González Macroeconomics Reading Group - UC3M January 2017

Introduction How important is financial development on economic development? First some warming up... The Role of Intermediaries: 1. Receive money from savers. 2. Collect and analyze information before investing into a business. 3. If they proceed to invest, decide how much and in what terms. 4. After investment, monitor firms activity to ensure savers best interest is protected. Efficiency of financial intermediation: measured by interest-rate spreads. Difference between the internal rate of return on investment in firms and the rate of return on saving received by savers.

Introduction Why do we care about efficiency of financial intermediation in a country? Negative relationship between interest spreads and capita-to-output ratio. J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 195 Fig. 1. Interest-rate spreads and capital-to-gdp ratios for the United States and Taiwan, 1970 2005. Data sources for all figures are discussed in Appendix A. information costs of policing investments, and the costs of misappropriation of savers funds by management, unions, and

Introduction Cross-country negative relationship between interest-rate spreads and GDP per capita. 196 J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Fig. 2. The cross-country relationship among interest-rate spreads, capital-to-gdp ratios and GDPs per capita. The three letter country codes are taken from the International Organization for Standardization, ISO 3166-1 alpha-3.

Introduction Cross-country negative relationship between interest-rate spreads and TFP. Fig. 2. The cross-country relationship among interest-rate spreads, capital-to-gdp ratios and GDPs per capita. The three letter country codes are taken from the International Organization for Standardization, ISO 3166-1 alpha-3. Fig. 3. The cross-country relationship among interest-rate spreads, TFPs and GDPs per capita. The model is calibrated to match some stylized facts for the U.S. economy, specifically the firm-size distributions and interest-rate spreads for the years 1974 and 2004. It replicates these facts very well. The improvement in financial sector productivity Greenwood et required al. (2013) to duplicate these Quantifying facts also theappears Impacttoofbe Financial reasonable; Development it does this with little change in the capital-to-

Introduction Aim of the paper Investigate quantitatively the impact of financial development on economic development using a costly state verification model. Two agents: Firms and Financial Intermediaries. Firms can lie about their true productivity. Probability of detecting malfeasance depends on: Labor used in monitoring activity chosen by financial intermediaries. Technology used in the financial sector z. Firms face a distribution of returns. Financial Theory of Firm size!

The Model Firms Firms Produce with CRS technology Aggregate productivity component: x o = xθ i k α i l 1 α i (1) Firms idiosyncratic type: τ = {θ 1, θ 2 } with θ 1 < θ 2 P(θ = θ 1 ) = π 1 ; P(θ = θ 2 ) = π 2 = 1 π 1 Firms type is common knowledge Realization of θ is private information.

The Model Firms Timing R(θ i, w)k max{xθ i ki α l 1 α l i wl i } (2) i 1. Raise k funds from financial intermediary before knowing realization of θ. 2. After observing θ, hire labor l i 3. Report to financial intermediary its θ...but they can lie! r i (τ) = R(θ i, w) = α(1 α) 1 α (1 α) α w α (xθ i ) 1/α > 0 (3)

The Model Financial Intermediaries Financial Intermediaries Competitive intermediation: intermediaries earn 0 profits. Raise funds from consumers and lend them to firms. Cannot costlessly observe θ (nor l or o) Only observe the report of the firm θ j Need to employ lmj to monitor the firm.

The Model Financial Intermediaries Probability that a firm is caught cheating: P ij (l mj, k, z) P ij (l mj, k, z) = { 1 1 < 1 for a report θ ɛ(z/k) ψ (l mj ) γ j θ i 0for a report θ j = θ i Parameters governing 0 < γ < ψ < 1 l mj is the labor allocated to monitoring activity. z is productivity in monitoring activity. k is the size of the loan. Diminishing returns to scale in monitoring!

The Model Financial Intermediaries subject to v max k,l m1 π 2 [1 P 21 (l m1, k, z)] [r 2 (τ) r 1 (τ)]k] (4) [π 1 r 1 (τ)+π 2 r 2 (τ)]k π 2 [1 P 21 (l m1, k, z)] [r 2 (τ) r 1 (τ)]k] π 1 wl m1 rk 0 }{{} Exp. return from type 2: p 2 (5) Maximand (4) are the rents that firms obtain. Equation (5) is the 0 profit condition for the intermediary. Incentive compatible contract [1 P 21 (l m1, k, z)] [r 2 (τ) r 1 (τ)]k] r 2 (τ)k p 2 (6)

The Model Stationary Equilibrium Firms are distributed over productivities according to F (x, y) = Pr(θ 1 x, θ 2 y) (7) π 1 and π 2 are constant across firms. Firms receiving funding in equilibrium will be A(w) {τ : π 1 r 1 (τ) + π 2 r 2 (τ) r > 0} (8) Labor market clearing condition [π 1 l 1 (θ 1, θ 2 ) + π 2 l 1 (θ 1, θ 2 ) + π 1 l m1 (θ 1, θ 2 )] df (θ 1, θ 2 ) = 1 A(w) (9)

The Model Stationary Equilibrium Definition 1: Stationary Competitive Equilibrium Set the steady-state cost of capital at r. A stationary competitive equilibrium is described by a set of labor allocations, l i and l m1, a loan size, k, and a value, v, for each firm, a set of active firms, A(w), and a wage rate, w, such that: 1. The loan, k, offered by the intermediary maximizes the value of a firm, v, in line with (4), given the prices r and w. The intermediary hires labor for monitoring in the amount l m1, as also specified by (4). 2. A firm is offered a loan if and only if it lies in the active set, A(w), as defined by Eq. (8). 3. A firm hires labor, l i, to maximize its profits in accordance with (2), given wages, w, and the size of the loan, k, offered by the intermediary. 4. The wage rate, w, is determined so that the labor market clears, in accordance with (9).

Calibration Fitting the model to the US economy Select 11 parameters with 18 data targets PARAM = (x, z, ɛ, ψ, γ, µ θ2, σθ 2 1, σθ 2 2, ρ) }{{} p where firm level ln(tfp) distributed according to a bivariate truncated normal. N(µ θ1, µ θ2, σ 2 θ 1, σ 2 θ 2, ρ) Targets: Size distribution of firms in 1974 and 2004. Constraints o [π 1 o 1 (θ 1, θ 2 ) + π 2 o 1 (θ 1, θ 2 )] df (θ 1, θ 2 ) (10) A(w) s w A(w) π 1l m1 (θ 1, θ 2 )df (θ 1, θ 2 ) A(w) k(θ 1, θ 2 )df (θ 1, θ 2 ) (11)

Calibration Cross-country calibration Mapping from the stationary equilibrium (x74 US, zus (o, s) = O(x, z; p) (x, z) = O 1 (o, s; p) (12) 74 ) = O 1 (o US 74, sus (x04 US, zus 04 ) = O 1 (o US 04, sus 74 ; p), 04 ; p) For each country j: back (x j, z j ) using (o j, s j ) from (12), with p = p USA. Parameter Values Matching of the Model

γ = 0.57 Pr of detection, exp on labor Calibrated to fit targets µ θ1 = 1.0 Mean of ln(θ1) Normalization µ θ2 = 2.26 Mean of ln(θ2) Calibrated to fit targets Results σ The US, balanced 2 θ1 growth = 0.70 Variance of ln(θ1) Calibrated to fit targets σ 2 = 0.27 Variance of ln(θ2) Calibrated to fit targets θ2 ρ = 0.80 Correlation ln(θ1) and ln(θ2) Calibrated to fit targets x1974 = 0.54, z1974 = 10.76 TFP s, 1974 Calibrated to fit targets x2004 = 0.77, z2004 = 26.44 TFP s, 2004 Calibrated to fit targets Table 2 The U.S. economy. Data Model 1974 Spread, s 3.07% 3.07% GDP (per capita), o $22,352 $22,352 Capital-to-output ratio (indexed), k/o 1.00 1.00 TFP 6.17 2004 Spread, s 2.62% 2.62% GDP (per capita), o $41,208 $41,208 Capital-to-output ratio (indexed), k/o 1.02 1.09 TFP 8.92 2004 Counterfactual, z US 2004 = zus 1974 Spread, s 2.62 3.93 GDP (per capita), o $41,208 $34,530 Capital-to-output ratio (indexed), k/o 1.02 0.87 TFP 8.59 Yearly growth in financial productivity 2.58% about 10 percent of TFP growth is due to improvement in financial intermediation. The financial system actually becomes adragondevelopmentwhenz is not allowed to increase. Wages rise as the rest of the economy develops. This makes monitoring more expensive; therefore, less monitoring will be done. As a consequence, interest rates rise and the economy s capital-to-output ratio drops. With no improvement in the financial system, the firm-size distribution actually moves over time in a direction (rightward) that is opposite to that shown in the data (leftward); this can be seen by comparing the lower Greenwood two panels et al. (2013) of Fig. 5. WhenQuantifying thereisnotechnological Impact of progress Financial the Development financialsector, therearemoresmallinefficient

Results Taiwan, unbalanced growth J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 205 Table 3 The Taiwan economy. Data Model 1974 Productivity, industrial x1974 = 0.14 Productivity, financial z1974 = 0.40 Spread, s 5.41% 5.41% GDP (per capita), o $2211 $2211 Capital-to-output (indexed), k/o 1.00 1.00 TFP 1.55 2004 Productivity, industrial x2004 = 0.35 Productivity, financial z2004 = 15.64 Spread, s 1.96% 1.96% GDP (per capita), o $13,924 $13,924 Capital-to-output (indexed), k/o 1.85 1.76 TFP 4.20 2004 Counterfactual, z T 2004 = zt 1974 Spread, s 1.96% 10.43% GDP (per capita), o $13,924 $6176 Capital-to-output (indexed), k/o 1.85 0.62 TFP 3.57 Yearly growth in financial productivity 9.90% How important was financial development for Taiwan s economic development? To answer this question, compute the model s solution for 2004 assuming there had been no financial development; that is, set z T 2004 = zt 1974.Almost45percent of Taiwan s 6.3 percent annual rate of growth between 1974 and 2004 can be attributed to financial development; it also accounts Greenwood foret 16 al. percent (2013) of the growth Quantifying in Taiwanese the Impact TFP. Taiwan of Financial had almost Development a 10 percent annual increase in the productivity

Results Cross-Country Results Obtain Best Practices in the World. x = max{x j } and z = max{z j } x = x US and z = z LUX (13) Gain from adopting best financial practices. O(x j, z) O(x j, z j ) Gain from adopting best financial and industrial practices. O( x, z) O(x j, z j ) Percentage of gap closed by adopting best financial practices. 100 [O(x j, z) O(x j, z j )]/[O( x, z) O(x j, z j )] Some Robustness on Imputed z

Results Cross-Country Results J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 207 Fig. 7. Cross-country results showing the impact of a move to financial best practice on GDP per worker, the output gap, and TFP the model. Therefore, given the higher wages, monitoring will be more expensive in the United States. For both countries to have the same interest-rate spread, efficiency in the U.S. financial sector must be higher. Before proceeding on to a discussion of the importance of financial Quantifying development forthe economic Impact development, of Financial note that the Development findings do not change much if the model is

Results The Importance of Financial Development J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Table 5 Worldwide move to best financial practice, z. Increase in world output (per worker), % 53.3 Reduction in gap between actual and potential world output, % 30.8 Increase in world TFP, % 13.5 Fall in dispersion of ln(output) across countries, % 22.8 Fall in (pop-wghtd) mean of (cap-wghtd) distortion, % 14.7 Fall in (pop-wghtd) mean dispersion of (cap-wghtd) distortion, % 9.5

Results The Importance of Financial Development 208 J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Distortion caused by informational frictions Table 5 Worldwide move to best financial practice, z. d = π 1 r 1 + π 2 r 2 r (14) Increase in world output (per worker), % 53.3 Reduction in gap between actual and potential world output, % 30.8 Increase in world TFP, % 13.5 Fall in dispersion of ln(output) across countries, % 22.8 Fall in (pop-wghtd) mean of (cap-wghtd) distortion, % 14.7 Fall in (pop-wghtd) mean dispersion of (cap-wghtd) distortion, % 9.5 Fig. 8. The distribution of distortions, d = π1r1 + π2r2 r, acrossestablishmentsforluxembourganduganda themodel. For most countries the shortfall in output is accounted for by a low level of TFP the nonfinancial sector. A more detailed (Capital-weighted) breakdown of the cross-countrymean results is presented level in Table of9 indistortion Appendix A. is 21% (2%) and Therefore, the importance of financial intermediation for economic development depends on one s outlook. World output std. deviation of 18% (1.2%) for Uganda (Luxembourg). would rise by 53 percent by moving all countries to the best financial practice (see Table 5). This is a sizable gain. Still, it would close only 31 percent of the gap between actual and potential world output. Dispersion in cross-country output would fall by about 15 percentage points from 77 percent to 62 percent. Financial development explains about 23 percent of cross-country dispersion in output by this metric. Restuccia and Rogerson (2008) started a strand in the literature about the importance of idiosyncratic distortions that create heterogeneity Quantifying in the prices thefaced Impact by individual of Financial producers. Although Development they do not identify the sources of those dis-

Results The Importance of Financial Development Model predicts larger firms should be found in countries with more developed financial systems. ln(size) = CONSTANT + η SPREAD + ι CONTROLS (15) J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 209 Table 6 Cross-country firm-size regressions. Data Model Interest-rate spread coefficient, η 22.4 16.6 Standard error for η 2.35 6.55 Number of country observations 27 27 R 2 0.80 0.53 On this point, imagine running a regression of the following form for both the data and the model: ln(size) Robustness = constant Analysis + η to spread Alternative + ι controls. Matching Strategies Firm size in the data is measured by average annual sales per firm (in U.S. dollars) for the top 100 firms, as taken from Beck et al. (2006). Fortheanalogueinthemodel,simplyuseacountry sgdpdividedbythemeasureofactiveseta to obtain output per firm. [Once again the data for interest-rate spreads are obtained from Beck et al. (2000, 2001).] Controls are added for a country s GDP and population in the regression for the data, while for the model they are added just for GDP. Greenwood 14 The same et al. list (2013) of countries isquantifying used for both the the Impact data and of Financial model. Development

Conclusions Use a costly state verification model to assess the importance of financial development on economic development. Introduces two unique features: Odds of detecting malfeasance depends on lmj (chosen endogenously) and z. There is an economywide distribution across firms over firm-specific distributions. Financial theory of firm size. Find that financial intermediation is important for economic development. World output could increase by 53% if all countries adopted the best financial practioce in the world. Still, this accounts only for 31% of the gap Huge differences in productivity of non-financial sector!

Thank you!

APPENDIX

Parameter Values 204 J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Table 1 Parameter values. Parameter Definition Basis α = 0.35 Capital s share of income Standard δ = 0.06 Depreciation rate Standard r = 0.03 Return to savers Siegel (1992) ϵ = 32.57 Pr of detection, constant Normalization ψ = 0.96 Pr of detection, exp on capital Calibrated to fit targets γ = 0.57 Pr of detection, exp on labor Calibrated to fit targets µ θ1 = 1.0 Mean of ln(θ1) Normalization µ θ2 = 2.26 Mean of ln(θ2) Calibrated to fit targets σ 2 = 0.70 θ1 Variance of ln(θ1) Calibrated to fit targets σ 2 = 0.27 Variance of ln(θ2) Calibrated to fit targets θ2 ρ = 0.80 Correlation ln(θ1) and ln(θ2) Calibrated to fit targets x1974 = 0.54, z1974 = 10.76 TFP s, 1974 Calibrated to fit targets x2004 = 0.77, z2004 = 26.44 TFP s, 2004 Calibrated to fit targets Back Table 2 The U.S. economy. Data Model 1974 Spread, s 3.07% 3.07% GDP (per capita), o $22,352 $22,352 Capital-to-output ratio (indexed), k/o 1.00 1.00 TFP 6.17 2004 Spread, s Quantifying the Impact of Financial 2.62% Development 2.62%

Matching of the Model J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 203 Fig. 5. Firm-size distributions, 1974 and 2004 U.S. data and model. subject to ( x US 74, ) Back ( zus 74 = O 1 o US 74, sus 74 ;p), (7) and ( US US ) ( Quantifying ) the Impact of Financial Development 1 US US

Some Robustness on Imputed z 206 J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Table 4 Cross-country evidence. k/o z with Beck et al. (2000, 2001) k/lm with Beck et al. (2000, 2001) Corr(model, data) 0.62 0.80 0.82 Fig. 6. The relationship between imputed ln(z) on the one hand, and measures of information technology, human capital, overhead costs to assets and the rule of law, on the other. Back panel) and the maturity of its legal system (lower-right panel). These three factors should make intermediation more efficient for the reasons discussed in Section 3. Indeed, Fig. 6 (lower-left panel) also illustrates how the ratio of overhead cost to assets, a measure of efficiency, declines with the constructed ln(z). Anothermeasureoffinancialefficiency for the model is k/lm; thiswasdiscussedearlier.thistoocorrelateswellwiththeindependentbeck et al. (2000, 2001) measure of financial intermediation. As can be seen, Quantifying the capital-to-output theratios Impact predictedof by the Financial model are positively Development associated with those in the data. The

210 J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Robustness Analysis to Alternative Matching Strategies What if internal funds do not require monitoring, and this source of finance is sizeable? Match k/o instead of s Financial sector is not competitive in less-developed countries There should be no correlation between overhead costs and interest rate spreads. Use overhead costs φ as proxy for for efficiency in financial sector. Fig. 9. The right panel shows the relationship between interest-rate spreads and capital-to-output ratios for the data and model. The left panel plots the relationship between interest-rate spreads and overhead costs for the data. Table 7 Worldwide move to best financial practice, z. Matching methodology s k/o φ Increase in world output (per worker), % 53.25 48.26 52.12 Reduction in gap between actual and potential world output, % 30.80 25.60 37.01 Increase in world TFP, % 13.47 14.28 13.10 Fall in dispersion of ln(output) across countries, % 22.83 32.82 13.82 with financial constraints may have difficulty accounting for the observed pattern of capital output ratios and interest-rate spreads (Fig. 9). Such models often emphasize a different mechanism: the impact that financing constraints have on TFP, through Backthe selection of entrepreneurs, as opposed to the impact on the aggregate capital-to-output ratio. One could also argue that the financial sector is not competitive in less-developed countries and that this accounts for their high interest-rate spreads. Suppose that the financial sector is monopolized in some countries. Intuitively, one may expect that monopolies might charge borrowers higher interest rates on loans and offer depositors lower rates on

Robustness Analysis to Alternative Matching Strategies 210 J. Greenwood et al. / Review of Economic Dynamics 16 (2013) 194 215 Fig. 9. The right panel shows the relationship between interest-rate spreads and capital-to-output ratios for the data and model. The left panel plots the relationship between interest-rate spreads and overhead costs for the data. Back Table 7 Worldwide move to best financial practice, z. Matching methodology s k/o φ Increase in world output (per worker), % 53.25 48.26 52.12 Reduction in gap between actual and potential world output, % 30.80 25.60 37.01 Increase in world TFP, % 13.47 14.28 13.10 Fall in dispersion Quantifying of ln(output) the acrossimpact countries, % of Financial Development 22.83 32.82 13.82