A Framework for Thinking about Technology Adoption Eric Verhoogen - - PowerPoint PPT Presentation

A Framework for Thinking about Technology Adoption

Eric Verhoogen

• Sept. 10, 2019

Introduction Framework Recent Research Conclusion

Introduction

◮ There is wide agreement that adoption of new and better technologies is central to the process of development.

Introduction Framework Recent Research Conclusion

Introduction

◮ There is wide agreement that adoption of new and better technologies is central to the process of development. ◮ Many technologies have already been developed in richer countries.

◮ Gerschenkron (1962): “advantages of backwardness.”

Introduction Framework Recent Research Conclusion

Introduction

◮ There is wide agreement that adoption of new and better technologies is central to the process of development. ◮ Many technologies have already been developed in richer countries.

◮ Gerschenkron (1962): “advantages of backwardness.”

◮ But for many countries, these advantages have remained elusive.

Introduction Framework Recent Research Conclusion

Introduction

◮ There is wide agreement that adoption of new and better technologies is central to the process of development. ◮ Many technologies have already been developed in richer countries.

◮ Gerschenkron (1962): “advantages of backwardness.”

◮ But for many countries, these advantages have remained elusive. ◮ What is getting in the way? A first-order question for development.

Introduction Framework Recent Research Conclusion

Introduction (cont.)

◮ In this talk I’ll present a framework for thinking about research on technology adoption by firms, aiming to provide:

◮ A way of interpreting existing studies. ◮ Suggestions for interesting directions for future research.

Introduction Framework Recent Research Conclusion

Introduction (cont.)

◮ In this talk I’ll present a framework for thinking about research on technology adoption by firms, aiming to provide:

◮ A way of interpreting existing studies. ◮ Suggestions for interesting directions for future research.

◮ I’m focusing on the determinants of technology adoption, not the consquences of technology adoption for other outcomes.

Introduction Framework Recent Research Conclusion

Introduction (cont.)

◮ In this talk I’ll present a framework for thinking about research on technology adoption by firms, aiming to provide:

◮ A way of interpreting existing studies. ◮ Suggestions for interesting directions for future research.

◮ I’m focusing on the determinants of technology adoption, not the consquences of technology adoption for other outcomes. ◮ Apology in advance to the non-economists for some mathematical notation. I’ll also try to explain in words.

Introduction Framework Recent Research Conclusion

A Framework

◮ Suppose products can be produced by various techniques, each characterized by a production function: Yijkt = Fijk(

⇀

Mijkt, λijkt)

◮ i = firm ◮ j = product ◮ k = “technique”. Think: machine + set of inputs. ◮

⇀

Mijkt = vector of amounts of inputs

◮ λijkt = “capability” in product-technique jk. Think:

entrepreneurial ability or know-how.

Introduction Framework Recent Research Conclusion

A Framework

◮ Suppose products can be produced by various techniques, each characterized by a production function: Yijkt = Fijk(

⇀

Mijkt, λijkt)

◮ i = firm ◮ j = product ◮ k = “technique”. Think: machine + set of inputs. ◮

⇀

Mijkt = vector of amounts of inputs

◮ λijkt = “capability” in product-technique jk. Think:

entrepreneurial ability or know-how.

◮ Firms face demand functions for products and supply functions for inputs: Pijt = Dijt(Yijt,

⇀

Yi,−jt, Zyt)

⇀

Wijkt = W (

⇀

Mijkt,

⇀

Mi,−jkt, Zmt)

Introduction Framework Recent Research Conclusion

A Framework (cont.)

◮ The firm’s present discounted value of profits is:

Πit =

∞

• t=0

δt   

• j∈J∗

it

• PijtFijk∗(

⇀

Mijk∗t, λijk∗t) −

⇀

W ′

ijk∗t

⇀

Mijk∗t − fijk∗t − fijt

• − fit

  

where δ is a discount factor and fijkt, fijt, fit are fixed costs.

Introduction Framework Recent Research Conclusion

A Framework (cont.)

◮ The firm’s present discounted value of profits is:

Πit =

∞

• t=0

δt   

• j∈J∗

it

• PijtFijk∗(

⇀

Mijk∗t, λijk∗t) −

⇀

W ′

ijk∗t

⇀

Mijk∗t − fijk∗t − fijt

• − fit

  

where δ is a discount factor and fijkt, fijt, fit are fixed costs. ◮ The firm’s problem is to choose the set of products to produce, J∗

it, the technique for each product, k∗ j , and amounts

• f inputs,

⇀

Mijkt, to maximize Πit.

Introduction Framework Recent Research Conclusion

A Framework (cont.)

◮ The firm’s present discounted value of profits is:

Πit =

∞

• t=0

δt   

• j∈J∗

it

• PijtFijk∗(

⇀

Mijk∗t, λijk∗t) −

⇀

W ′

ijk∗t

⇀

Mijk∗t − fijk∗t − fijt

• − fit

  

where δ is a discount factor and fijkt, fijt, fit are fixed costs. ◮ The firm’s problem is to choose the set of products to produce, J∗

it, the technique for each product, k∗ j , and amounts

• f inputs,

⇀

Mijkt, to maximize Πit. ◮ (Could incorporate investments in capabilities, knowledge: IitΛ, IitJ, IitK.)

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption

◮ Suppose that two techniques, kL and kH, use the same set of inputs and kH is “better” than kL for firms in developed countries, in the sense that: Πit(kH; {λijkt}, {Pijt}, {

⇀

Wijkt}) > Πit(kL; {λijkt}, {Pijt}, {

⇀

Wijkt}) assuming other choices are made optimally.

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption

◮ Suppose that two techniques, kL and kH, use the same set of inputs and kH is “better” than kL for firms in developed countries, in the sense that: Πit(kH; {λijkt}, {Pijt}, {

⇀

Wijkt}) > Πit(kL; {λijkt}, {Pijt}, {

⇀

Wijkt}) assuming other choices are made optimally. ◮ Caveat: This does not necessarily mean that kH is more technically efficient, in the sense of getting more physical

• utput from the same physical inputs.

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption

◮ Suppose that two techniques, kL and kH, use the same set of inputs and kH is “better” than kL for firms in developed countries, in the sense that: Πit(kH; {λijkt}, {Pijt}, {

⇀

Wijkt}) > Πit(kL; {λijkt}, {Pijt}, {

⇀

Wijkt}) assuming other choices are made optimally. ◮ Caveat: This does not necessarily mean that kH is more technically efficient, in the sense of getting more physical

• utput from the same physical inputs.

◮ kH, kL may be used for different products (e.g. quality

varieties), hence physical output not directly comparable.

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption

◮ Suppose that two techniques, kL and kH, use the same set of inputs and kH is “better” than kL for firms in developed countries, in the sense that: Πit(kH; {λijkt}, {Pijt}, {

⇀

Wijkt}) > Πit(kL; {λijkt}, {Pijt}, {

⇀

Wijkt}) assuming other choices are made optimally. ◮ Caveat: This does not necessarily mean that kH is more technically efficient, in the sense of getting more physical

• utput from the same physical inputs.

◮ kH, kL may be used for different products (e.g. quality

varieties), hence physical output not directly comparable.

◮ kH, kL may entail different labor costs. A technically efficient

technology may increase workers’ bargaining power, not be the more profitable (Stole and Zwiebel, 1996)).

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption (cont.)

◮ Key question: Why might a firm in a developing country not adopt kH?

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption (cont.)

◮ Key question: Why might a firm in a developing country not adopt kH? ◮ I’m going to classify reasons in three broad categories:

• 1. Internal-to-the-firm
• 2. Input-side
• 3. Output-side

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption (cont.)

◮ Key question: Why might a firm in a developing country not adopt kH? ◮ I’m going to classify reasons in three broad categories:

• 1. Internal-to-the-firm
• 2. Input-side
• 3. Output-side

◮ The reasons in turn point to factors that might promote adoption.

Introduction Framework Recent Research Conclusion

Reasons for Non-Adoption (cont.)

◮ Key question: Why might a firm in a developing country not adopt kH? ◮ I’m going to classify reasons in three broad categories:

• 1. Internal-to-the-firm
• 2. Input-side
• 3. Output-side

◮ The reasons in turn point to factors that might promote adoption. ◮ Then I’ll come back and discuss two studies in more detail.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons

◮ The firm is not profit-maximizing.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons

◮ The firm is not profit-maximizing.

◮ Leibenstein (1966): “X-inefficiency.”

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons

◮ The firm is not profit-maximizing.

◮ Leibenstein (1966): “X-inefficiency.” ◮ May be true, but we should be very careful about jumping to

this conclusion.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons

◮ The firm is not profit-maximizing.

◮ Leibenstein (1966): “X-inefficiency.” ◮ May be true, but we should be very careful about jumping to

this conclusion.

◮ If firms appear not to be optimizing, perhaps we have not fully

understood the problem they are solving.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons

◮ The firm is not profit-maximizing.

◮ Leibenstein (1966): “X-inefficiency.” ◮ May be true, but we should be very careful about jumping to

this conclusion.

◮ If firms appear not to be optimizing, perhaps we have not fully

understood the problem they are solving.

◮ The firm does not know about kH.

◮ Even in the age of Google, firms are sometimes simply unaware

• f existing technologies.

◮ A firm may have uncertainty about how well a technology

works, even if operated correctly.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons

◮ The firm is not profit-maximizing.

◮ Leibenstein (1966): “X-inefficiency.” ◮ May be true, but we should be very careful about jumping to

this conclusion.

◮ If firms appear not to be optimizing, perhaps we have not fully

understood the problem they are solving.

◮ The firm does not know about kH.

◮ Even in the age of Google, firms are sometimes simply unaware

• f existing technologies.

◮ A firm may have uncertainty about how well a technology

works, even if operated correctly.

◮ The firm lacks capability for implementing kH.

◮ It appears in many cases that capabilities have to be

“home-grown”, and at a cost (Gibbons, 2010).

◮ If a firm lacks the required capabilities for technique kH, it may

not be profitable to adopt it.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons (cont.)

◮ First reason (X-inefficiency) points to increasing competition as a factor driving adoption.

◮ In some cases, it seems clear that competition raised

productivity (Schmitz, 2005; Das et al., 2013).

◮ Little work on competition and technology adoption per se.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons (cont.)

◮ First reason (X-inefficiency) points to increasing competition as a factor driving adoption.

◮ In some cases, it seems clear that competition raised

productivity (Schmitz, 2005; Das et al., 2013).

◮ Little work on competition and technology adoption per se.

◮ Second and third reasons (lack of knowledge, capabilities) point to learning as a driver of adoption.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons (cont.)

◮ First reason (X-inefficiency) points to increasing competition as a factor driving adoption.

◮ In some cases, it seems clear that competition raised

productivity (Schmitz, 2005; Das et al., 2013).

◮ Little work on competition and technology adoption per se.

◮ Second and third reasons (lack of knowledge, capabilities) point to learning as a driver of adoption.

◮ Learning can take various forms:

◮ An increase in capabilities, {λijkt} ◮ An enlargement of set of techniques a firm knows about, Kijt. ◮ An enlargement of set of products a firm knows about, Jit.

Introduction Framework Recent Research Conclusion

Internal-to-the-Firm Reasons (cont.)

◮ First reason (X-inefficiency) points to increasing competition as a factor driving adoption.

◮ In some cases, it seems clear that competition raised

productivity (Schmitz, 2005; Das et al., 2013).

◮ Little work on competition and technology adoption per se.

◮ Second and third reasons (lack of knowledge, capabilities) point to learning as a driver of adoption.

◮ Learning can take various forms:

◮ An increase in capabilities, {λijkt} ◮ An enlargement of set of techniques a firm knows about, Kijt. ◮ An enlargement of set of products a firm knows about, Jit.

◮ And can happen through various channels:

◮ From peers (Cai and Szeidl, 2017; Hardy and McCasland,

2016)

◮ From customers or suppliers. ◮ Through worker flows. ◮ Consulting (Bloom et al., 2013; Bruhn et al., 2018) ◮ ...

Introduction Framework Recent Research Conclusion

Input-Side Reasons

◮ Developing-country firms face different input-supply functions, {

⇀

Wijkt}.

◮ Newer, more advanced techniques often require highly skilled

workers, high-quality inputs.

◮ These may be expensive, or unavailable.

Introduction Framework Recent Research Conclusion

Input-Side Reasons

◮ Developing-country firms face different input-supply functions, {

⇀

Wijkt}.

◮ Newer, more advanced techniques often require highly skilled

workers, high-quality inputs.

◮ These may be expensive, or unavailable.

◮ Points to changes in input prices/availability as a driver of adoption.

Introduction Framework Recent Research Conclusion

Input-Side Reasons

◮ Developing-country firms face different input-supply functions, {

⇀

Wijkt}.

◮ Newer, more advanced techniques often require highly skilled

workers, high-quality inputs.

◮ These may be expensive, or unavailable.

◮ Points to changes in input prices/availability as a driver of adoption.

◮ A number of studies show that reduction of import tariffs leads

to product upgrading (Bas and Strauss-Kahn, 2015).

Introduction Framework Recent Research Conclusion

Input-Side Reasons

◮ Developing-country firms face different input-supply functions, {

⇀

Wijkt}.

◮ Newer, more advanced techniques often require highly skilled

workers, high-quality inputs.

◮ These may be expensive, or unavailable.

◮ Points to changes in input prices/availability as a driver of adoption.

◮ A number of studies show that reduction of import tariffs leads

to product upgrading (Bas and Strauss-Kahn, 2015).

◮ Some work on labor supply and mechanization in agriculture

(Hornbeck and Naidu, 2014).

Introduction Framework Recent Research Conclusion

Input-Side Reasons

◮ Developing-country firms face different input-supply functions, {

⇀

Wijkt}.

◮ Newer, more advanced techniques often require highly skilled

workers, high-quality inputs.

◮ These may be expensive, or unavailable.

◮ Points to changes in input prices/availability as a driver of adoption.

◮ A number of studies show that reduction of import tariffs leads

to product upgrading (Bas and Strauss-Kahn, 2015).

◮ Some work on labor supply and mechanization in agriculture

(Hornbeck and Naidu, 2014).

◮ Little evidence on input supply and technology adoption per se

in LDC firms.

Introduction Framework Recent Research Conclusion

Output-Side Reasons

◮ Developing-country firms face different product-demand functions, {Pijt}.

Introduction Framework Recent Research Conclusion

Output-Side Reasons

◮ Developing-country firms face different product-demand functions, {Pijt}.

◮ Different techniques, e.g. kH and kL, may be applicable in the

production of different (but possibly similar) products.

◮ e.g. different quality varieties.

Introduction Framework Recent Research Conclusion

Output-Side Reasons

◮ Developing-country firms face different product-demand functions, {Pijt}.

◮ Different techniques, e.g. kH and kL, may be applicable in the

production of different (but possibly similar) products.

◮ e.g. different quality varieties.

◮ Customer demand for different products may differ between

developed/developing countries.

◮ e.g. Willingness to pay for quality greater in developed

countries.

◮ Customers may have preferences over techniques used in

production as well as product characteristics.

Introduction Framework Recent Research Conclusion

Output-Side Reasons

◮ Developing-country firms face different product-demand functions, {Pijt}.

◮ Different techniques, e.g. kH and kL, may be applicable in the

production of different (but possibly similar) products.

◮ e.g. different quality varieties.

◮ Customer demand for different products may differ between

developed/developing countries.

◮ e.g. Willingness to pay for quality greater in developed

countries.

◮ Customers may have preferences over techniques used in

production as well as product characteristics. ◮ Market-size often smaller in developing countries.

◮ May be difficult to achieve scale to pay for fixed costs.

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

◮ Exports ↑ ⇒ quality ↑ (Verhoogen, 2008; Atkin et al., 2017a).

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

◮ Exports ↑ ⇒ quality ↑ (Verhoogen, 2008; Atkin et al., 2017a). ◮ Selling to local MNCs ↑ ⇒ quality ↑ (Alfaro-Urena et al., 2019)

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

◮ Exports ↑ ⇒ quality ↑ (Verhoogen, 2008; Atkin et al., 2017a). ◮ Selling to local MNCs ↑ ⇒ quality ↑ (Alfaro-Urena et al., 2019) ◮ Exports, selling to local MNCs ↑ ⇒ working conditions ↑

(Tanaka, forthcoming; Boudreau, 2019).

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

◮ Exports ↑ ⇒ quality ↑ (Verhoogen, 2008; Atkin et al., 2017a). ◮ Selling to local MNCs ↑ ⇒ quality ↑ (Alfaro-Urena et al., 2019) ◮ Exports, selling to local MNCs ↑ ⇒ working conditions ↑

(Tanaka, forthcoming; Boudreau, 2019).

◮ Exports ↑ ⇒ scale ↑ (Bustos, 2011; Lileeva and Trefler, 2010).

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

◮ Exports ↑ ⇒ quality ↑ (Verhoogen, 2008; Atkin et al., 2017a). ◮ Selling to local MNCs ↑ ⇒ quality ↑ (Alfaro-Urena et al., 2019) ◮ Exports, selling to local MNCs ↑ ⇒ working conditions ↑

(Tanaka, forthcoming; Boudreau, 2019).

◮ Exports ↑ ⇒ scale ↑ (Bustos, 2011; Lileeva and Trefler, 2010). ◮ Competition ↑ ⇒ X-inefficiency ↓

Introduction Framework Recent Research Conclusion

Output-Side Reasons (cont.)

◮ Potential drivers:

◮ Exports ↑ ⇒ quality ↑ (Verhoogen, 2008; Atkin et al., 2017a). ◮ Selling to local MNCs ↑ ⇒ quality ↑ (Alfaro-Urena et al., 2019) ◮ Exports, selling to local MNCs ↑ ⇒ working conditions ↑

(Tanaka, forthcoming; Boudreau, 2019).

◮ Exports ↑ ⇒ scale ↑ (Bustos, 2011; Lileeva and Trefler, 2010). ◮ Competition ↑ ⇒ X-inefficiency ↓ ◮ Again, not much direct evidence on technology adoption.

Introduction Framework Recent Research Conclusion

Consulting Experiment

◮ Bloom, Eifert, Mahajan, McKenzie and Roberts (2013) ◮ Management practices can be thought of as a technology.

◮ “Modern management is a technology that diffuses slowly

between firms.” (Bloom et al., 2011)

Introduction Framework Recent Research Conclusion

Consulting Experiment

◮ Bloom, Eifert, Mahajan, McKenzie and Roberts (2013) ◮ Management practices can be thought of as a technology.

◮ “Modern management is a technology that diffuses slowly

between firms.” (Bloom et al., 2011)

◮ Useful to study not only because they seem to matter a lot, but also because practices are used across a wide range of firms.

◮ Many other technologies need to be studied within narrow

industries.

Introduction Framework Recent Research Conclusion

Consulting Experiment

◮ Bloom, Eifert, Mahajan, McKenzie and Roberts (2013) ◮ Management practices can be thought of as a technology.

◮ “Modern management is a technology that diffuses slowly

between firms.” (Bloom et al., 2011)

◮ Useful to study not only because they seem to matter a lot, but also because practices are used across a wide range of firms.

◮ Many other technologies need to be studied within narrow

industries.

◮ This study randomly allocated intensive Accenture consulting services over 17 Indian textile firms.

◮ One-month diagnostic phase (all firms) ◮ Four-month implementation phase (treatment only) ◮ Market value of services ∼ $250k. ◮ Tracked 38 management practices (regular maintenance,

tracking inventory, performance-based pay systems).

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Firms had not implemented basic management practices (e.g. labelling inventory).

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

FIGURE V The Adoption of Key Textile Management Practices over Time

◮ Consulting was successful in getting firms to adopt management practices.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Quality defects declined sharply.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

TABLE II THE IMPACT OF MODERN MANAGEMENT PRACTICES ON PLANT PERFORMANCE

• Notes. All regressions use a full set of plant and calendar week dummies. Standard errors are bootstrap clustered at the firm level. Intervention is a plant level dummy equal to

• f the implementation phase for all treatment plants. Cumulative treatment is the cumulative weeks of treatment since the beginning of the implementation phase in each plant (0

• available. Small sample robustness implements two different procedures (described in greater detail in the Appendix and Online Appendix B) to address issues of plant hetero-

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Study shows convincingly that management consulting can raise firm performance.

◮ That’s already a significant achievement, rightfully influential.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Study shows convincingly that management consulting can raise firm performance.

◮ That’s already a significant achievement, rightfully influential.

◮ Study has been interpreted as showing:

◮ Firms are systematically making mistakes by not adopting the

38 practices.

◮ The 38 practices themselves unambiguously raise firm

performance.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Study shows convincingly that management consulting can raise firm performance.

◮ That’s already a significant achievement, rightfully influential.

◮ Study has been interpreted as showing:

◮ Firms are systematically making mistakes by not adopting the

38 practices.

◮ The 38 practices themselves unambiguously raise firm

performance.

◮ Some words of caution:

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Study shows convincingly that management consulting can raise firm performance.

◮ That’s already a significant achievement, rightfully influential.

◮ Study has been interpreted as showing:

◮ Firms are systematically making mistakes by not adopting the

38 practices.

◮ The 38 practices themselves unambiguously raise firm

performance.

◮ Some words of caution:

◮ Adoption of modern management may require capabilities that

firms did not have and that are costly to acquire.

◮ $250k market value in consulting yielded about $300k increase

in profit per year, noisily measured.

◮ Experiment paid cost for firms, but not clear they were

making a mistake by not paying it on their own.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ Study shows convincingly that management consulting can raise firm performance.

◮ That’s already a significant achievement, rightfully influential.

◮ Study has been interpreted as showing:

◮ Firms are systematically making mistakes by not adopting the

38 practices.

◮ The 38 practices themselves unambiguously raise firm

performance.

◮ Some words of caution:

◮ Adoption of modern management may require capabilities that

firms did not have and that are costly to acquire.

◮ $250k market value in consulting yielded about $300k increase

in profit per year, noisily measured.

◮ Experiment paid cost for firms, but not clear they were

making a mistake by not paying it on their own. ◮ Question: Are effects due to 38 management practices

themselves or to other effects of consulting?

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ My interpretation:

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ My interpretation:

◮ Experiment raised capabilities of firms, which allowed them to

implement modern management practices.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ My interpretation:

◮ Experiment raised capabilities of firms, which allowed them to

implement modern management practices.

◮ Some of the practices do seem like “no-brainers” that would

be advantageous for all firms in all contexts (e.g. tracking inventories). But others are less obvious (e.g. performance pay).

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ My interpretation:

◮ Experiment raised capabilities of firms, which allowed them to

implement modern management practices.

◮ Some of the practices do seem like “no-brainers” that would

be advantageous for all firms in all contexts (e.g. tracking inventories). But others are less obvious (e.g. performance pay).

◮ We need to evaluate setting by setting whether particular

practices are indeed advantageous, worth promoting.

Introduction Framework Recent Research Conclusion

Consulting Experiment (cont.)

◮ My interpretation:

◮ Experiment raised capabilities of firms, which allowed them to

implement modern management practices.

◮ Some of the practices do seem like “no-brainers” that would

be advantageous for all firms in all contexts (e.g. tracking inventories). But others are less obvious (e.g. performance pay).

◮ We need to evaluate setting by setting whether particular

practices are indeed advantageous, worth promoting.

◮ The fundamental problem seems to be that firms lack

capabilities, not that they have failed to adopt a particular set

• f practices.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment

◮ Atkin, Chaudhry, Chaudry, Khandelwal and Verhoogen (2017b). ◮ Through a series of fortuitous events, we came up with a technology that seemed to be as close to a no-brainer as one could hope for.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment

◮ Atkin, Chaudhry, Chaudry, Khandelwal and Verhoogen (2017b). ◮ Through a series of fortuitous events, we came up with a technology that seemed to be as close to a no-brainer as one could hope for. ◮ Randomly allocated it to 35 of 135 soccer-ball producers in Sialkot, Pakistan.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment

◮ Material (rexine) is by far the most expensive input.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

Standard “buckyball” design: 20 hexagons, 12 pentagons. For standard ball, almost all firms use 2-hexagon and 2- pentagon “flush” dies.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

Hexagons tessellate. ∼ 8% of rexine wasted.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

Pentagons don’t. ∼ 20-24% of rexine wasted.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

In a YouTube video of a Chinese factory producing the Adidas Jabulani ball, I noticed a different layout of pentagons.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

I could also have gone to: G. Kuperberg and W. Kuperberg, “Double-Lattice Packings of Convex Bodies in the Plane,” Discrete & Computational Geometry, 5: 389-397, 1990.

Double-Lattice Packings

• f Convex Bodies

in the Plane 393 inscribed in Ko, and the proof is complete. However, it can be noticed now that the minimality of the area of q implies that the length of one of the sides of q actually equals one-half of the length of Ko in the direction of that side. Therefore Ko actually touches a translate of itself, and Case II is not possible at all. [] Remark 1. If K is not strictly convex, the conclusion of the above theorem does not necessarily hold. However, in this case there exists a double-lattice packing with maximum density which is generated by a minimum-area extensive parallelogram inscribed in K. This can be obtained by approximating K with a sequence of strictly convex bodies K, and then selecting a convergent subsequence

• f double-lattice packings.

Remark 2. Theorem 1 and the above remark yield an algorithm for finding a maximum density double-lattice packing with copies of K which goes as follows. For any diameter d of K, find a pair of chords parallel to d, each of length equal to one-half of the length of d. These two chords define a parallelogram q(d) inscribed in K, which turns out to be extensive (see Lemma 1 of the following section). Now vary d and find a critical position of d = do such that q(do) is of minimum area. This minimum-area extensive parallelogram generates a maximum density double-lattice packing with copies of K. In general, locating the critical diameter do may be a problem, but in many special cases, as in the following examples, the diameter do is easy to find. Examples. An application of the algorithm described in Remark 2 to the case when K is a regular pentagon results in a double-lattice packing of density (5-x/5)/3 =0.92131..., shown in Fig. 7. This packing may have the maximum

• Fig. 7. Maximum

density double-lattice packing with regular pentagons.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

Or the Wikipedia Pentagons page:

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

Annalisa Guzzini (an architect, my wife) and I developed a blueprint based on optimal packing.

• ◮ 44mm-edge pentagons: ∼250 with old die vs. 272 with ours.

◮ 43.5mm-edge pentagons: ∼258 vs. 280.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

Introduction Framework Recent Research Conclusion

Net Benefits of Adoption

10th 25th 50th 75th 90th mean net variable cost reduction (%) 0.42 0.61 0.82 1.09 1.47 0.89 (0.11) (0.10) (0.09) (0.19) (0.27) (0.11) % net variable cost/avg % profit rate 4.55 6.82 10.63 16.56 24.42 13.07 (1.05) (1.13) (1.60) (2.35) (4.15) (1.79) total cost savings per month (Rs 000s) 3.66 9.82 41.35 135.92 397.95 137.77 (0.99) (2.33) (9.43) (36.39) (130.62) (31.68) total cost savings per cutter per month (Rs 000s) 2.75 6.47 14.91 33.83 63.61 27.31 (0.83) (1.33) (2.43) (6.28) (14.02) (5.04) days to recover fixed costs 10.28 19.11 43.03 100.86 247.53 168.80 (2.23) (3.66) (7.37) (21.74) (76.42) (84.72) days to recover fixed costs (no die) 5.34 9.92 22.34 52.37 128.53 87.64 (1.16) (1.90) (3.83) (11.29) (39.68) (43.99)

◮ We estimate that 50% of tech drop firms would recover fixed costs in 23 days or less, 75% in 53 days or less.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

◮ We did a second experiment.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

◮ We did a second experiment.

◮ Offered bonus of one month’s pay to show owners that dies

work.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

◮ We did a second experiment.

◮ Offered bonus of one month’s pay to show owners that dies

work.

◮ Workers accepted and about 50% of affected firms adopted.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

◮ We did a second experiment.

◮ Offered bonus of one month’s pay to show owners that dies

work.

◮ Workers accepted and about 50% of affected firms adopted.

◮ We argue that the results are not due simply to us subsidizing the fixed costs of adoption.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

◮ We did a second experiment.

◮ Offered bonus of one month’s pay to show owners that dies

work.

◮ Workers accepted and about 50% of affected firms adopted.

◮ We argue that the results are not due simply to us subsidizing the fixed costs of adoption.

◮ Hard to reconcile with both initial non-adoption and adoption

in response to small incentives.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Quick summary:

◮ We gave out the new dies to a random subset of firms. ◮ A few adopted, most did not. ◮ Cutters didn’t like the die.

◮ They are paid piece rates, with no incentive to reduce waste,

and the new die was slowing them down.

◮ They told owners it didn’t work.

◮ We did a second experiment.

◮ Offered bonus of one month’s pay to show owners that dies

work.

◮ Workers accepted and about 50% of affected firms adopted.

◮ We argue that the results are not due simply to us subsidizing the fixed costs of adoption.

◮ Hard to reconcile with both initial non-adoption and adoption

in response to small incentives.

◮ It appears that second experiment induced information flow in

firm.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Interpretation:

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Interpretation:

◮ Conflict of interest within the firm can prevent adoption of a

“no-brainer” technology.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Interpretation:

◮ Conflict of interest within the firm can prevent adoption of a

“no-brainer” technology.

◮ The conflict of interest can be thought of as a lack of

• rganizational capability.

Introduction Framework Recent Research Conclusion

Soccer-ball Experiment (cont.)

◮ Interpretation:

◮ Conflict of interest within the firm can prevent adoption of a

“no-brainer” technology.

◮ The conflict of interest can be thought of as a lack of

• rganizational capability.

◮ Reinforces view of Bloom et al. (2013) that “information

failures” explain lack of adoption.

Introduction Framework Recent Research Conclusion

Conclusion

◮ The mere fact that Πit(kH) > Πit(kL) in developed countries does not imply that Πit(kH) > Πit(kL) in developing countries.

Introduction Framework Recent Research Conclusion

Conclusion

◮ The mere fact that Πit(kH) > Πit(kL) in developed countries does not imply that Πit(kH) > Πit(kL) in developing countries.

◮ Developing-country firms have different capabilities, {λijkt},

and face different product-demand and input-supply functions {Pijt}, {

⇀

Wijkt}.

Introduction Framework Recent Research Conclusion

Conclusion

◮ The mere fact that Πit(kH) > Πit(kL) in developed countries does not imply that Πit(kH) > Πit(kL) in developing countries.

◮ Developing-country firms have different capabilities, {λijkt},

and face different product-demand and input-supply functions {Pijt}, {

⇀

Wijkt}.

◮ Statement applies to management practices as well as to

machines.

Introduction Framework Recent Research Conclusion

Conclusion

◮ The mere fact that Πit(kH) > Πit(kL) in developed countries does not imply that Πit(kH) > Πit(kL) in developing countries.

◮ Developing-country firms have different capabilities, {λijkt},

and face different product-demand and input-supply functions {Pijt}, {

⇀

Wijkt}.

◮ Statement applies to management practices as well as to

machines.

◮ That said, it does appear that there are cases where more technically efficient technologies are not adopted, e.g.:

◮ Keeping track of inventories. ◮ Offset die.

Introduction Framework Recent Research Conclusion

Conclusion

◮ The mere fact that Πit(kH) > Πit(kL) in developed countries does not imply that Πit(kH) > Πit(kL) in developing countries.

◮ Developing-country firms have different capabilities, {λijkt},

and face different product-demand and input-supply functions {Pijt}, {

⇀

Wijkt}.

◮ Statement applies to management practices as well as to

machines.

◮ That said, it does appear that there are cases where more technically efficient technologies are not adopted, e.g.:

◮ Keeping track of inventories. ◮ Offset die.

◮ We have some results on what is getting in the way, e.g.:

◮ Lack of manager know-how. ◮ Organizational barriers to information flows.

Introduction Framework Recent Research Conclusion

Conclusion (cont.)

◮ But the literature on technology adoption in non-agricultural firms in developing countries (with direct observation of adoption) is still pretty wide open.

◮ Little evidence on input-side drivers of adoption. ◮ Little evidence on output-side drivers of adoption. ◮ More to do on learning, capabilities, internal-to-the-firm

drivers.

Introduction Framework Recent Research Conclusion

Conclusion (cont.)

◮ But the literature on technology adoption in non-agricultural firms in developing countries (with direct observation of adoption) is still pretty wide open.

◮ Little evidence on input-side drivers of adoption. ◮ Little evidence on output-side drivers of adoption. ◮ More to do on learning, capabilities, internal-to-the-firm

drivers.

◮ We’ve got out work cut out for us!

Introduction Framework Recent Research Conclusion

References I

Alfaro-Urena, Alfonso, Isabela Manelici, and Jose P. Vasquez, “The Effects of Joining Multinational Supply Chains: New Evidence from Firm-to-Firm Linkages,” 2019. Unpub. paper. Atkin, David, Amit K. Khandelwal, and Adam Osman, “Exporting and Firm Performance: Evidence from a Randomized Trial,” Quarterly Journal of Economics, 2017, 132 (2), 551–615. , Azam Chaudhry, Shamyla Chaudry, Amit K. Khandelwal, and Eric Verhoogen, “Organizational Barriers to Technology Adoption: Evidence from Soccer-Ball Producers in Pakistan,” Quarterly Journal of Economics, 2017, 132 (3), 1101–1164. Bas, Maria and Vanessa Strauss-Kahn, “Input-Trade Liberalization, Export Prices and Quality Upgrading,” Journal

• f International Economics, 2015, 95 (2), 250–262.

Bloom, Nicholas, Benn Eifert, Aprajit Mahajan, David McKenzie, and John Roberts, “Does Management Matter? Evidence from India,” 2011. NBER Working Paper No. 16658. , , , , and , “Does Management Matter? Evidence from India,” Quarterly Journal of Economics, February 2013, 128 (1), 1–51. Boudreau, Laura, “Multinational Enforcement of Labor Law: Experimental Evidence from Bangladesh’s Apparel Sector,” 2019. Unpub. paper. Bruhn, Miriam, Dean Karlan, and Antoinette Schoar, “The Impact of Consulting Services on Small and Medium Enterprises: Evidence from a Randomized Trial in Mexico,” Journal of Political Economy, 2018, 126 (2), 635–687. Bustos, Paula, “Trade Liberalization, Exports and Technology Upgrading: Evidence on the Impact of MERCOSUR

• n Argentinian Firms,” American Economic Review, 2011, 101 (1), 304–340.

Cai, Jing and Adam Szeidl, “Interfirm Relationships and Business Performance,” The Quarterly journal of economics, 2017, 133 (3), 1229–1282.

Introduction Framework Recent Research Conclusion

References II

Das, Sanghamitra, Kala Krishna, Sergey Lychagin, and Rohini Somanathan, “Back on the rails: competition and productivity in state-owned industry,” American Economic Journal: Applied Economics, 2013, 5 (1), 136–162. Gerschenkron, Alexander, Economic Backwardness in Historical Perspective: A Book of Essays, Harvard University Press, 1962. Gibbons, Robert, “Inside Organizations: Pricing, Politics, and Path Dependence,” Annual Review of Economics, 2010, 2 (1), 337–365. Hardy, Morgan and Jamie McCasland, “It Takes Two: Experimental Evidence on the Determinants of Technology Diffusion,” 2016. Unpub. paper, University of British Columbia. Hornbeck, R. and S. Naidu, “When the Levee Breaks: Black Migration and Economic Development in the American South,” American Economic Review, 2014, 104 (3), 963–990. Kuperberg, G. and W. Kuperberg, “Double-Lattice Packings of Convex Bodies in the Plane,” Discrete & Computational Geometry, 1990, 5, 389–397. Leibenstein, Harvey, “Allocative Efficiency vs. X-efficiency,” American Economic Review, 1966, 56, 24–31. Lileeva, Alla and Daniel Trefler, “Improved Access to Foreign Markets Raises Plant-Level Productivity ... For Some Plants,” Quarterly Journal of Economics, August 2010, 125 (3), 1051–1099. Schmitz, James A., “What Determines Productivity? Lessons from the Dramatic Recovery of the U.S. and Canadian Iron Ore Industries Following Their Early 1980s Crisis,” Journal of Political Economy, 2005, 113 (3),

• pp. 582–625.

Stole, Lars A. and Jeffrey Zwiebel, “Intra-Firm Bargaining under Non-Binding Contracts,” The Review of Economic Studies, 1996, 63 (3), pp. 375–410.

Introduction Framework Recent Research Conclusion

References III

Tanaka, Mari, “Exporting Sweatshops? Evidence from Myanmar,” Review of Economics and Statistics, forthcoming. Verhoogen, Eric, “Trade, Quality Upgrading, and Wage Inequality in the Mexican Manufacturing Sector,” Quarterly Journal of Economics, 2008, 123 (2), 489–530.

Thank you!