pith. sign in

arxiv: 1906.08359 · v1 · pith:3PAS5VUXnew · submitted 2019-06-19 · 📊 stat.AP · stat.ME

An axiomatic nonparametric production function estimator: Modeling production in Japan's cardboard industry

Daisuke Yagi , Yining Chen , Andrew L. Johnson , Hiroshi Morita This is my paper

Pith reviewed 2026-05-25 19:35 UTC · model grok-4.3

classification 📊 stat.AP stat.ME
keywords production function estimationnonparametric regressionshape constraintsRegular Ultra Passum lawJapanese cardboard industryproductivity decompositionscale efficiencycapital-to-labor ratio
0
0 comments X

The pith

Shape-constrained nonparametric regression shows most productive scale size depends on the capital-to-labor ratio in Japan's cardboard industry.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a production function estimator by translating the Regular Ultra Passum law and convex non-homothetic input isoquants into shape constraints for nonparametric regression. Applied to 1997-2007 panel data from the Japanese corrugated cardboard industry, the estimator indicates that most productive scale size is a function of the capital-to-labor ratio. Largest firms operate near the largest most productive scale size linked to high capital-to-labor ratios. Productivity growth is recovered from the model residuals and decomposed into scale, input mix, and unexplained components.

Core claim

By enforcing the Regular Ultra Passum law and convex non-homothetic input isoquants as shape constraints within nonparametric regression, the axiomatic estimator recovers the underlying production function for the Japanese cardboard industry, revealing that most productive scale size varies with the capital-to-labor ratio and that the largest firms sit close to the largest such scale at high capital intensity.

What carries the argument

Shape-constrained nonparametric regression that imposes the Regular Ultra Passum law and convex non-homothetic input isoquants.

If this is right

  • Most productive scale size increases with the capital-to-labor ratio.
  • Largest firms operate near the most productive scale size associated with high capital-to-labor ratios.
  • Productivity differences across firms and periods can be attributed separately to scale, input mix, and residual effects.
  • Productivity growth over time can be measured directly from the residuals of the fitted axiomatic model.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same shape-constrained approach could be applied to other industries to test whether optimal scale depends on input ratios in a similar way.
  • The decomposition supplies managers with concrete levers—adjusting scale or input proportions—to address measured productivity shortfalls.
  • If the axioms hold more broadly, the method offers a way to estimate production relationships without choosing a parametric functional form in advance.

Load-bearing premise

The true production function for the Japanese cardboard industry obeys the Regular Ultra Passum law and exhibits convex non-homothetic input isoquants.

What would settle it

Finding that the estimated production function under these shape constraints produces isoquants that are non-convex or that returns to scale violate the Regular Ultra Passum law in the observed data would falsify the approach.

Figures

Figures reproduced from arXiv: 1906.08359 by Andrew L. Johnson, Daisuke Yagi, Hiroshi Morita, Yining Chen.

Figure 1
Figure 1. Figure 1: Production functions satisfying both the RUP law a [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Input isoquants satisfying input convexity. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of functional estimates. origin. Therefore, our observed input vector Xj = (Xj1, . . . , Xjd) ′ in spherical coordinates system (rj , φj ) = (rj , φj,1, . . . , φj,d−1) is defined as: rj = q X2 j1 + . . . + X2 jd φj,1 = arccos Xj1 q X2 j1 + . . . + X2 jd φj,2 = arccos q Xj2 X2 j2 + . . . + X2 jd . . . φj,d−1 = arccos q Xj,d−1 X2 j,d−1 + X2 jd , (5) where rj is the radial distance from the orig… view at source ↗
Figure 4
Figure 4. Figure 4: Isoquant estimation 3.4.2 S-shape estimation To estimate the S-shape functions on rays from the origin, we begin by project all observations {Xj , yj} n j=1 to each ray from the origin θ (r) . We can either project the observations directly onto the rays, or use the estimated isoquants from the previous step to project the observations. For the second approach, in short, we find the level of an isoquant to… view at source ↗
Figure 5
Figure 5. Figure 5: S-shape estimation 3.4.3 Computing functional estimates at a given input vector The last step of Algorithm 1 obtains the functional estimates ˆg(x) at any given value of input vector x, and computes the MSE against observations {Xj , yj} n j=1. First we compute the weighted average of the two closest isoquants which sandwich the observed input Xj . The details are given in Appendix B.3.1. Second, we assign… view at source ↗
Figure 6
Figure 6. Figure 6: Gap between convex isoquant and S-shape estimates [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Estimation results on the testing sets with the par [PITH_FULL_IMAGE:figures/full_fig_p025_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Estimation results on the testing sets with the non [PITH_FULL_IMAGE:figures/full_fig_p026_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Estimation results on the testing sets with the non [PITH_FULL_IMAGE:figures/full_fig_p028_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Price deflator (Base year = 2000) 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Year 70.0% 80.0% 90.0% 100.0% 110.0% 120.0% Labor growth Average of 0-25 percentile of amount produced Average of 25-50 percentile of amount produced Average of 50-75 percentile of amount produced Average of 75-100 percentile of amount produced [PITH_FULL_IMAGE:figures/full_fig_p031_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Percentage change of quartile mean of labor [PITH_FULL_IMAGE:figures/full_fig_p031_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Percentage change of quartile mean of capital [PITH_FULL_IMAGE:figures/full_fig_p032_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Percentage change quartile mean of value added [PITH_FULL_IMAGE:figures/full_fig_p032_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Centroid of each group estimated by K-means clustering 0 2 4 6 8 10 12 14 16 Labor 0 2 4 6 8 10 12 14 16 Capital Most Productive Scale Size (a) Estimated input isoquants (b) Estimated production function [PITH_FULL_IMAGE:figures/full_fig_p035_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Estimated results of the corrugated cardboard in [PITH_FULL_IMAGE:figures/full_fig_p035_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Percentage change of quartile mean of productivi [PITH_FULL_IMAGE:figures/full_fig_p036_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Percentage change of quartile mean of input ratio [PITH_FULL_IMAGE:figures/full_fig_p037_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Productivity decomposition of firms belong to the [PITH_FULL_IMAGE:figures/full_fig_p038_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: Productivity decomposition of firms belong to the [PITH_FULL_IMAGE:figures/full_fig_p039_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: Noise which is orthogonal to the true isoquant [PITH_FULL_IMAGE:figures/full_fig_p046_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Estimated isoquant by CNLS, Directional CNLS and [PITH_FULL_IMAGE:figures/full_fig_p046_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: Procedures of the projection of the observation i [PITH_FULL_IMAGE:figures/full_fig_p048_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: Kernel weight in the S-shape estimation Here we define a distance measure in angles by their L2 distance (in the d−1 Euclidean space), i.e. D(φ1, φ2) = kφ1 − φ2k2. For each ray from the origin θ (r) , we solve the following quadratic programming problem: min a,b,g (r) ∗ X G g=1 Xn j=1  y˜j −  a (r) g + b (r) g  R (r) j − r (r) g 2 K  D(φj ,θ (r) ) ω  k  R (r) j −r (r) g h(r)  subject to a (r) g … view at source ↗
Figure 24
Figure 24. Figure 24: How to obtain intersecting points r (i)(r) is resolved through this step. Here, we describe the mathematical formulation. We start from redefining the evaluation points on a ray, θ (r) as r (r) g ∈ {r r 1 , . . . , rr G} ∀g = 1, . . . , G r (r) g (i) ∈ {r (1)(r) , . . . , r(I)(r) } ∀i = 1, . . . , I r (r) g ′ ∈ {r r 1 , . . . , rr G} ∪ {r (1)(r) , . . . , r(I)(r) } ∀g ′ = 1, . . . , G′ (27) where G′ = G +… view at source ↗
Figure 25
Figure 25. Figure 25: Modification of S-shape estimates min a˜ (r) g w S · 1 R · G X R r=1 X G g=1  a˜ (r) g − a (r) g 2 + w I · 1 R · I X R r=1 X I i=1  a˜ (r) g (i) − y (i) 2 subject to a˜ (r) g+2 − a˜ (r) g+1 r (r) g+2 − r (r) g+1 ≥ a˜ (r) g+1 − a˜ (r) g r (r) g+1 − r (r) g ∀r and ∀g = 1, . . . , g (r) ∗ − 2 a˜ (r) g+2 − a˜ (r) g+1 r (r) g+2 − r (r) g+1 ≤ a˜ (r) g+1 − a˜ (r) g r (r) g+1 − r (r) g ∀r and ∀g = g (r) ∗ − 2… view at source ↗
Figure 26
Figure 26. Figure 26: Estimation results on the testing set for the isoq [PITH_FULL_IMAGE:figures/full_fig_p057_26.png] view at source ↗
Figure 27
Figure 27. Figure 27: The elasticity of scale 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 (a) Production function 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 (b) First derivative 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 (c) Second derivative [PITH_FULL_IMAGE:figures/full_fig_p069_27.png] view at source ↗
Figure 28
Figure 28. Figure 28: Production function and its derivatives 69 [PITH_FULL_IMAGE:figures/full_fig_p069_28.png] view at source ↗
Figure 29
Figure 29. Figure 29: Productivity decomposition (Group–1) [PITH_FULL_IMAGE:figures/full_fig_p073_29.png] view at source ↗
Figure 30
Figure 30. Figure 30: Productivity decomposition (Group–2) 73 [PITH_FULL_IMAGE:figures/full_fig_p073_30.png] view at source ↗
Figure 31
Figure 31. Figure 31: Productivity decomposition (Group–3) [PITH_FULL_IMAGE:figures/full_fig_p074_31.png] view at source ↗
Figure 32
Figure 32. Figure 32: Productivity decomposition (Group–4) 74 [PITH_FULL_IMAGE:figures/full_fig_p074_32.png] view at source ↗
Figure 33
Figure 33. Figure 33: Productivity decomposition (Group–5) [PITH_FULL_IMAGE:figures/full_fig_p075_33.png] view at source ↗
Figure 34
Figure 34. Figure 34: Productivity decomposition (Group–6) 75 [PITH_FULL_IMAGE:figures/full_fig_p075_34.png] view at source ↗
Figure 35
Figure 35. Figure 35: Productivity decomposition (Group–7) [PITH_FULL_IMAGE:figures/full_fig_p076_35.png] view at source ↗
Figure 36
Figure 36. Figure 36: Productivity decomposition (Group–8) 76 [PITH_FULL_IMAGE:figures/full_fig_p076_36.png] view at source ↗
Figure 37
Figure 37. Figure 37: Productivity decomposition (Group–9) [PITH_FULL_IMAGE:figures/full_fig_p077_37.png] view at source ↗
Figure 38
Figure 38. Figure 38: Productivity decomposition (Group–10) 77 [PITH_FULL_IMAGE:figures/full_fig_p077_38.png] view at source ↗
Figure 39
Figure 39. Figure 39: Productivity decomposition (Group–11) [PITH_FULL_IMAGE:figures/full_fig_p078_39.png] view at source ↗
Figure 40
Figure 40. Figure 40: Productivity decomposition (Group–12) 78 [PITH_FULL_IMAGE:figures/full_fig_p078_40.png] view at source ↗
read the original abstract

We develop a new approach to estimate a production function based on the economic axioms of the Regular Ultra Passum law and convex non-homothetic input isoquants. Central to the development of our estimator is stating the axioms as shape constraints and using shape constrained nonparametric regression methods. We implement this approach using data from the Japanese corrugated cardboard industry from 1997-2007. Using this new approach, we find most productive scale size is a function of the capital-to-labor ratio and the largest firms operate close to the largest most productive scale size associated with a high capital-to-labor ratio. We measure the productivity growth across the panel periods based on the residuals from our axiomatic model. We also decompose productivity into scale, input mix, and unexplained effects to clarify the sources the productivity differences and provide managers guidance to make firms more productive.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper develops a nonparametric estimator for production functions by translating the Regular Ultra Passum law and convex non-homothetic input isoquants into shape constraints for regression. It applies the estimator to Japanese corrugated cardboard industry panel data (1997-2007) and reports that most productive scale size is a function of the capital-to-labor ratio, that the largest firms operate near the largest most productive scale size associated with high capital-to-labor ratios, and that productivity growth (measured from residuals) can be decomposed into scale, input-mix, and unexplained components.

Significance. If the axioms hold for the industry, the approach offers a way to embed specific economic shape restrictions directly into nonparametric estimation, avoiding both fully parametric forms and completely unconstrained fits. The application supplies industry-specific evidence on scale economies varying with input ratios and a decomposition that isolates managerially relevant sources of productivity differences. The use of a decade-long panel on a narrowly defined industry is a concrete strength.

major comments (2)
  1. [Abstract] Abstract: the central empirical claims (MPS as a function of K/L, largest firms near high-K/L MPS, and the scale/input-mix/unexplained decomposition) rest on the shape-constrained estimator recovering the true production function. No derivation details, error analysis, or robustness checks are supplied, so it is impossible to assess whether the imposed constraints materially distort the recovered function when the axioms are only approximately satisfied.
  2. [Abstract] Abstract: no diagnostic is described that checks whether the Regular Ultra Passum law or convex non-homothetic isoquants bind on the data, whether unconstrained nonparametric fits violate the axioms by more than sampling error, or whether the reported MPS locations and residual-based productivity measures are sensitive to modest relaxations of either constraint. This validation step is load-bearing for all reported findings.
minor comments (1)
  1. The abstract does not state the exact definitions of capital, labor, and output or the source of the Japanese industry data; adding these would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and insightful comments. We address each major comment below, clarifying where the manuscript already provides relevant material and indicating revisions to strengthen validation of the estimator.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central empirical claims (MPS as a function of K/L, largest firms near high-K/L MPS, and the scale/input-mix/unexplained decomposition) rest on the shape-constrained estimator recovering the true production function. No derivation details, error analysis, or robustness checks are supplied, so it is impossible to assess whether the imposed constraints materially distort the recovered function when the axioms are only approximately satisfied.

    Authors: The translation of the Regular Ultra Passum law and convex non-homothetic isoquants into shape constraints is derived in Section 3, where the economic axioms are stated mathematically and imposed via inequality restrictions on the nonparametric regression. We agree, however, that the manuscript lacks explicit error analysis and robustness checks for cases where the axioms hold only approximately. In revision we will add simulation experiments that quantify estimator bias under controlled violations of the axioms and report sensitivity of the MPS and productivity decomposition to these perturbations. revision: yes

  2. Referee: [Abstract] Abstract: no diagnostic is described that checks whether the Regular Ultra Passum law or convex non-homothetic isoquants bind on the data, whether unconstrained nonparametric fits violate the axioms by more than sampling error, or whether the reported MPS locations and residual-based productivity measures are sensitive to modest relaxations of either constraint. This validation step is load-bearing for all reported findings.

    Authors: The current manuscript does not contain the requested diagnostics. We will therefore add, in a new subsection, (i) a comparison of the shape-constrained fit against an unconstrained nonparametric estimator to test whether violations exceed sampling variability, (ii) a check of whether the imposed constraints are binding at the estimated MPS points, and (iii) sensitivity results obtained by modestly relaxing each constraint in turn and recomputing the MPS schedule and productivity decomposition. revision: yes

Circularity Check

0 steps flagged

No circularity: estimator applies external axioms as shape constraints; results are data-driven outputs, not self-referential reductions.

full rationale

The derivation imposes the Regular Ultra Passum law and convex non-homothetic isoquants as shape constraints within standard nonparametric regression, then applies the resulting estimator to Japanese cardboard industry panel data. Central claims (MPS as function of K/L, firm locations relative to MPS, productivity decomposition) are direct outputs of this constrained fit rather than quantities defined by the fit itself or recovered via self-citation chains. No quoted step reduces a prediction to a fitted parameter by construction, and the axioms are presented as external economic primitives rather than derived from the estimator.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The estimator rests on two domain assumptions from economic theory that are imposed as shape constraints; no free parameters or invented entities are mentioned. The central claim depends on these axioms holding for the industry and on the nonparametric procedure correctly recovering the function under the constraints.

axioms (2)
  • domain assumption Regular Ultra Passum law
    Converted into a shape constraint that the estimated production function must satisfy.
  • domain assumption convex non-homothetic input isoquants
    Converted into a shape constraint that the estimated production function must satisfy.

pith-pipeline@v0.9.0 · 5679 in / 1376 out tokens · 29470 ms · 2026-05-25T19:35:10.310249+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages · 1 internal anchor

  1. [1]

    Afriat, S. N. (1972). Efficiency estimation of production functions. International Economic Review\/ 13\/ (3), 568--598

    work page 1972

  2. [2]

    Aksaray, G. and P. Thompson (2017). Density dependence of entrepreneurial dynamics: Competition, opportunity cost, or minimum efficient scale? Management Science\/ 64\/ (5), 2263--2274

    work page 2017

  3. [3]

    Barro, R. and X. Sala-i Martin (2004). Economic Growth . McGraw-Hill

    work page 2004

  4. [4]

    Baumol, W. J., J. C. Panzar, and R. D. Willig (1983). Contestable markets: An uprising in the theory of industry structure: Reply. The American Economic Review\/ 73\/ (3), 491--496

    work page 1983

  5. [5]

    Bogetoft, P. (1996). DEA on relaxed convexity assumptions. Management Science\/ 42\/ (3), 457--465

    work page 1996

  6. [6]

    Chambers, R. G., Y. Chung, and R. F \"a re (1998). Profit, directional distance functions, and nerlovian efficiency. Journal of Optimization Theory and Applications\/ 98\/ (2), 351--364

    work page 1998

  7. [7]

    Chen, Y. and R. J. Samworth (2016). Generalized additive and index models with shape constraints. Journal of the Royal Statistical Society: Series B (Statistical Methodology)\/ 78\/ (4), 729--754

    work page 2016

  8. [8]

    Chen, Y. and J. A. Wellner (2016). On convex least squares estimation when the truth is linear. Electronic Journal of Statistics\/ 10\/ (1), 171--209

    work page 2016

  9. [9]

    Christensen, L. R., D. W. Jorgenson, and L. J. Lau (1973). Transcendental logarithmic production frontiers. The Review of Economics and Statistics\/ 55\/ (1), 28--45

    work page 1973

  10. [10]

    Cleveland, W. S. (1979). Robust locally weighted regression and smoothing scatterplots. Journal of the American statistical association\/ 74\/ (368), 829--836

    work page 1979

  11. [11]

    Fan, Y. and E. Guerre (2016). Multivariate local polynomial estimators: Uniform boundary properties and asymptotic linear representation. In Essays in Honor of Aman Ullah , pp.\ 489--537. Emerald Group Publishing Limited

    work page 2016

  12. [12]

    F rsund, F. R. (1975). The homothetic production function. The Swedish Journal of Economics\/ 77\/ (2), 234--244

    work page 1975

  13. [13]

    F rsund, F. R. and L. Hjalmarsson (2004). Are all scales optimal in DEA ? theory and empirical evidence. Journal of Productivity Analysis\/ 21\/ (1), 25--48

    work page 2004

  14. [14]

    Haltiwanger, and C

    Foster, L., J. Haltiwanger, and C. Syverson (2016). The slow growth of new plants: Learning about demand? Economica\/ 83\/ (329), 91--129

    work page 2016

  15. [15]

    Frisch, R. (1964). Theory of production . Springer Science & Business Media

    work page 1964

  16. [16]

    Ginsberg, W. (1974). The multiplant firm with increasing returns to scale. Journal of Economic Theory\/ 9\/ (3), 283--292

    work page 1974

  17. [17]

    Haltiwanger, J., R. S. Jarmin, and J. Miranda (2013). Who creates jobs? small versus large versus young. Review of Economics and Statistics\/ 95\/ (2), 347--361

    work page 2013

  18. [18]

    Hwangbo, H., A. L. Johnson, and Y. Ding (2015). Power curve estimation: Functional estimation imposing the regular ultra passum law. SSRN working paper available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2621033

    work page 2015

  19. [19]

    Konishi, and Y

    Ichimura, H., Y. Konishi, and Y. Nishiyama (2011). An econometric analysis of firm specific productivities: Evidence from japanese plant level data. Discussion papers, Research Institute of Economy, Trade and Industry (RIETI)

    work page 2011

  20. [20]

    Iguchi, M. (2015). Growth strategy of cardboard industry: demand forecast and growth directions. Master's thesis, Waseda University (in Japanese)

    work page 2015

  21. [21]

    Kuosmanen, T. (2008). Representation theorem for convex nonparametric least squares. The Econometrics Journal\/ 11\/ (2), 308--325

    work page 2008

  22. [22]

    Johnson, and A

    Kuosmanen, T., A. Johnson, and A. Saastamoinen (2015). Stochastic nonparametric approach to efficiency analysis: A unified framework. In Data Envelopment Analysis , pp.\ 191--244. Springer

    work page 2015

  23. [23]

    Kuosmanen, T. and A. L. Johnson (2017). Modeling joint production of multiple outputs in stoned: Directional distance function approach. European Journal of Operational Research\/ 262\/ (2), 792--801

    work page 2017

  24. [24]

    Liao, X. and M. C. Meyer (2017). Change-point estimation using shape-restricted regression splines. Journal of Statistical Planning and Inference\/ 188 , 8--21

    work page 2017

  25. [25]

    Lim, E. and P. W. Glynn (2012). Consistency of multidimensional convex regression. Operations Research\/ 60\/ (1), 196--208

    work page 2012

  26. [26]

    Olesen, O. B. and J. Ruggiero (2014). Maintaining the regular ultra passum law in data envelopment analysis. European Journal of Operational Research\/ 235\/ (3), 798--809

    work page 2014

  27. [27]

    Petersen, N. C. (1990). Data envelopment analysis on a relaxed set of assumptions. Management Science\/ 36\/ (3), 305--314

    work page 1990

  28. [28]

    Seijo, E. and B. Sen (2011). Nonparametric least squares estimation of a multivariate convex regression function. The Annals of Statistics\/ 39\/ (3), 1633--1657

    work page 2011

  29. [29]

    Stone, C. J. (1977). Consistent nonparametric regression. The Annals of Statistics\/ 5\/ (4), 595--620

    work page 1977

  30. [30]

    Syverson, C. (2004). Product substitutability and productivity dispersion. The Review of Economics and Statistics\/ 86\/ (2), 534--550

    work page 2004

  31. [31]

    Syverson, C. (2011). What determines productivity? Journal of Economic literature\/ 49\/ (2), 326--365

    work page 2011

  32. [32]

    van der Vaart, A. and J. Wellner (1996). Weak Convergence and Empirical Processes: With Applications to Statistics . Springer Series in Statistics. Springer

    work page 1996

  33. [33]

    Varian, H. R. (1984). The nonparametric approach to production analysis. Econometrica\/ 52\/ (3), 579--597

    work page 1984

  34. [34]

    Yagi, D., Y. Chen, A. L. Johnson, and T. Kuosmanen (2018). Shape constrained kernel-weighted least squares: Application to production function estimation for chilean manufacturing industries. Accepted at the Journal of Business and Economic Statistics\/ , arXiv working paper available at: https://arxiv.org/abs/1604.06003

    work page internal anchor Pith review Pith/arXiv arXiv 2018