How Do Firms Grow? The Product Life Cycle Matters David Argente
Munseob Lee
University of Chicago
UC San Diego
Sara Moreira Northwestern University
Introduction I
How do firms grow? I
I
Firms grow I I I
I
Among many other stories, we focus on creative destruction inside the firm, the process by which new innovations replace older technologies
by increasing revenue from existing products by improving their own products by introducing new varieties and potentially displacing other firms’ products
Little is known empirically about the processes of creation and destruction of products, and how these processes differ over the firm’s lifecycle literature review 1
1. Data
1
What is a product?
I
A ”product” is uniquely identified by a 12-digit number called Universal Product Code (UPC), which is the finest level of disaggregation at the product level I I
Approximately 200 thousand on average every quarter Examples: a 31-ounce bag of Tide Pods Detergent.
Alternatively, a ”product” is a brand within a general type of product Approximately 50 thousand on average every quarter Examples: Tide Pods Detergent.
2
What is a product?
I
A “product” is uniquely identified by a 12-digit number called Universal Product Code (UPC), which is the finest level of disaggregation at the product level I I
I
Approximately 200 thousand on average every quarter Examples: a 31-ounce bag of Tide Pods Detergent.
Alternatively, a “product” is a brand within a general type of product I I
Approximately 50 thousand on average every quarter Examples: Tide Pods Detergent.
2
Baseline product-level dataset I
Source I I I
I
Nielsen Retail Measurement Services scanner dataset 2006-2015 40,000 food, drug and mass merchandising stores (90 retail chains) $220 billion of transactions/year, representing roughly 30 percent of total U.S. expenditures on food and beverages Report weekly sales and volume for every UPC with positive sales generated by point-of-sales systems. barcode datasets
I
Our product dataset I I I
Combines all sales at national and quarterly level Entry is the first quarter in which we observe sales of a product Exit is the first quarter after we last observe a product being sold.
3
Baseline product-level dataset: descriptive statistics
Average # of products
2007-2013
2007
2010
2013
222,105
211,101
214,001
252,189
Share of products by status
Entrants Exits
0.043 0.036
0.047 0.046
0.037 0.033
0.047 0.030
Share of products by revenue
[0,500[ [500,5000[ [5000,50000[ >=50000
0.610 0.230 0.136 0.024
0.626 0.223 0.128 0.023
0.605 0.232 0.138 0.025
0.615 0.228 0.135 0.022
4
Baseline firm-product dataset
I
Matching firm and products I
GS1 codes are part of the UPC code (first 6 or 10 digits of the code) upc structure
I
Our firm-product dataset I
I I
Our combined dataset allows us to identify the portfolio of products of each firm at a quarter level Firm Entry is first quarter of sales of the first product(s) by a firm Firm Exit is first quarter after we last observe the last product(s) being sold by a firm.
5
Baseline firm-product dataset: descriptive statistics
Average # of firms
2007-2013
2007
2010
2013
12,861
13,074
12,361
13,319
Share of firms by status
Entrants Exits
0.021 0.020
0.025 0.023
0.017 0.018
0.024 0.019
Share of firms by revenue
[0,104 [ [104 ,105 [ [105 ,106 [ >=106
0.462 0.262 0.182 0.093
0.480 0.256 0.177 0.087
0.455 0.265 0.185 0.095
0.462 0.264 0.180 0.094 top 10
6
2. Brief summary of our own previous work
6
Aggregate Creative Destruction 1/2 1. Product creation and destruction rates are remarkably large I
Over a typical twelve-month interval, about one in five products are created, and a comparable number is no longer available
2. Most reallocation of products occurs within the boundaries of the firm I
Entry and exit of firms have only a small contribution
3. Product entry and exit is strongly procyclical and declined by roughly 30 percent during the Great Recession I
This cyclical pattern is almost entirely explained by a decline in within firm reallocation.
7
Aggregate Creative Destruction 2/2
I
The creation and destruction of products includes products that are truly innovative and are related to the innovation efforts of firms
I
The creation and destruction of products has implications for the outcomes of firms: revenue growth, improvements in average product quality, and productivity growth
I
Our estimates suggest that the decline in product reallocation through these margins contributed importantly to the slow growth experienced after the Great Recession.
8
3. Stylized Facts on Product and Firm Lifecycle
8
Product Lifecycle: standard understanding
I
Harvard Business Review November 1965 Issue, T. Levitt
9
Revenue of product over the lifecycle: standard path
-.4
Revenue (log) -.2
0
.2
The revenue of products is mostly declining with age, with exception of the first 4-5 quarters.
-.6
I
1
2
3
4
5 6 Age in quarters
7
8
9
10
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
(Balanced Sample with duration +10 quarters) 10
Revenue of product over the lifecycle: standard path
Revenue (log) -.5 0
.5
... This empirical fact is robust to samples restricted to more long-lasting products
-1
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
(Balanced Sample with duration +16 quarters) 11
Revenue of product over the lifecycle: long lasting products
-1
Revenue (log) -.5
0
.5
... even to products that last at least 28 quarters
-1.5
I
4
8
12 16 Age in quarters
20
24
28
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
(Balanced Sample with duration +28 quarters) 12
Revenue of product over the lifecycle: by duration
Total Revenue (log) -15 -10 -5
0
... For short lasting products the revenue declines throughout all lifecycle until exit
-20
I
1
3 8 16
5 9 17
7
9 11 13 Age in quarters 10 18
11 19
12 20
15 13 21
17 14 >21
19 15
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
13
On duration...
0.25
Survival function 0.50
0.75
1.00
half of the products exit before they reach 16 quarters
0.00
I
4
8
12
16 20 24 Duration (in quarters)
28
32
36
40
Unconditional Kaplan-Meier Survival Function 14
On initial level ...
4 Initial Revenue (log) 2 3 1
I
Within very narrow defined types of products, there is substantial heterogeneity in the level of revenue that they generate. Short lasting products generate little revenue immediately at entry
0
I
4
8 Duration (in quarters)
12
16
Note: duration 1 normalized to equal 0; Conditional on module x quarter fixed-effects
15
Does the decline in revenue result from prices and/or quantity?
I
Marginal cost affects firm revenue through prices, while firm’s appeal affects firm revenue conditional on prices.
I
An advantage of our data is that we observe prices and quantities, and hence we are able to separate out cost and appeal as sources of dispersion in firm revenue.
16
Does the decline in revenue result from prices and/or quantity?
-.08
Price (log) -.06 -.04
-.02
0
Prices are declining throughout the entire lifecycle
-.1
I
1
2
3
4
5 6 Age in quarters
7
8
9
10
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
(Balanced Sample with duration +10 quarters) 17
Does the decline in revenue result from prices and/or quantity?
Price (log) -.2
-.1
0
Prices are declining for short and long-lasting products This is consistent with declining marginal costs (learning by doing?)
-.3
I
-.4
I
4 3 11
4 12
8 Age in quarters 5 13
6 14
7 15
12 8 16
16 9 >28
10 18
Does the decline in revenue result from prices and/or quantity?
-.4
Quantity (log) -.2 0
.2
Quantities are mostly declining with age, with exception of the first 4-5 quarters.
-.6
I
1
2
3
4
5 6 Age in quarters
7
8
9
10
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
(Balanced Sample with duration +10 quarters)
19
Does the decline in revenue result from prices and/or quantity?
0
Quantities are always declining for short-lasting products Consistent with larger appeal for new products (appeal declining with age)
Quantity (log) -4 -2
I
-6
I
4 3
4
8 Age in quarters 5
6
7
12 8
16 9
10
20
Does the decline in revenue result from prices and/or quantity?
-.5
0
.5
Decline in quantities is larger relatively to decline in prices.
-1
I
4 Price (log)
8 Age in quarters Quantity (log)
12
16 Revenue (log)
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
21
Implications for firm growth
Steady growth cannot result from within products growth, can come from: 1. Increasing the portfolio of products 2. Adding new products with larger duration 3. Adding new products with higher initial revenue
Note: Steady growth by single-product firms can only happen if replace product with decaying revenue by a new product with higher initial level of revenue.
22
Implications for firm growth: size of portfolio? The stock of products increases over the lifecycle of surviving firms
0
Total Number of Products (log) .1 .2 .3
.4
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
I
with a growth rate that declines with age 23
Implications for firm growth: size of portfolio The decline in the growth rate of size of the portfolio results from the reduction in the rate of new products
-.08
-.06
entry rate -.04
-.02
0
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
0
exit rate .01
.02
.03
... and the increase in the rate of product destruction
-.01
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
24
Implications for firm growth: initial revenue?
Total Revenue of New Products (log) 0 .5
1
The initial revenue of post-entry products is fairly constant
-.5
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
25
Implications for firm growth: heterogeneity? Revenue of improvements is stable Total Revenue of New Products Improvements (log) -.5 0 .5 1 1.5
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
Revenue of extensions declines the most Total Revenue of New Products Extensions (log) -1 -.5 0 .5 1
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
26
Implications for firm growth: heterogeneity?
Total Number of New products Improvements (log) -.4 -.2 0 .2 .4
4
8 Age in quarters
12
16
4
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
Total Number of New products Extensions (log) -1.5 -1 -.5 0 .5
Average Revenue of New Products Extensions (log) -4 -3 -2 -1 0 1
Revenue of extensions declines because of less products and products generating lower revenue
Total Revenue of New Products Extensions (log) -5 -4 -3 -2 -1 0
I
8 Age in quarters
Average Revenue of New Products Improvements (log) -.5 0 .5 1 1.5
Revenue of improvements is increasing (non-significant) because of the average revenue increases, while the number of products is stable
Total Revenue of New Products Improvements (log) -.5 0 .5 1 1.5
I
4
8 Age in quarters
12
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
16
4
8 Age in quarters
12
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
16
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
27
Implications for firm growth: heterogeneity
1 new products extensions, 0 new products improvements only -.4 -.3 -.2 -.1 0 .1
I
The likelihood of extending the portfolio as opposed of adding improvements declines with age
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
4
8 Age in quarters
12
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
16
Total Number of Departments (log) .02 .04 .06 0
0
0
Total Number of Groups (log) .05 .1 .15
Total Number of Modules (log) .05 .1 .15 .2
.2
.08
still the stock of modules, groups and departments increases with age
.25
I
4
8 Age in quarters
12
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
16
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
28
0
.2
Total Revenue (log) .4 .6
.8
Firm growth: revenue
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
29
Firm growth: price index
-.1
Index Quality (log) -.05
0
.05
Growth comes from increase in quantities not level of prices
-.15
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on cohort and main department x quarter fixed-effects
30
Firm growth: price index, role of entry, entry improvements, and exit Index Quality of Entry Products Improvements(log) -.4 -.2 0 .2
The relative of price of new products declines with age
-.4
Index Quality of Entry Products(log) -.2 0
.2
I
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
8 Age in quarters
12
16
Index Quality of Exit Products (log) -.5 0
.5
The relative of price of exits is stable
-1
I
4
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
4
8 Age in quarters
12
16
Note: age 1 normalized to equal 0; Conditional on firm and quarter fixed-effects
31
Firm growth: price index
I
The relative of price of new products declines with age
32
Next:
I
Model based dynamic decomposition framework that characterizes the contributions of the size of the product portfolio and its characteristics to firms’ revenue growth across their age distribution.
33
