2.5. Fundamentación legal
2.5.1. Constitucion de la Republica del Ecuador
In this section, I discuss some of the model’s implications. First of all, the model sheds light on the recent debate of the rise of aggregate markup in the economy. If I estimate the aggregate markup as the cost-weighted one in the model, the aggregate markup is virtually flat since the late 1980s. The same is true for the markup distribution. Consistent with data, a rise of returns to marketing cannot generate sizable changes in the markup distribution in the model.
Secondly, using cross-country data on marketing spending, Corrado and Hao (2014a) document a positive correlation between media advertising intensity and the level of economic development. Besides, for a given level of advertising intensity, there is a considerable variation of income per capita across countries. Our model provides a framework to understand these two phenomena.
Lastly, I show that the MPCR-markup elasticity only depends on the aggregate returns of marketing technology. Thus, using a fraction of marketing spending to measure MPCR would generate identical results in terms of level of the elasticity and its co-movement with aggregate marketing intensity.
1.7.1
Trend of market power in the data
In the model, aggregate markup that matters for resource allocation and pro- ductivity is production-cost weighted firm-level markup
µ=X k,n µk(Qn) W Lkp(Qn) W Lp Mk(Qn).
I then examine how a rise in returns to marketing affects aggregate markup and markup distribution in the economy.
Table 1.9 shows the aggregate markup and distribution of markup in the model. The rising returns to marketing alone cannot generate a sizable change in the markup distribution. Figure 1.10 shows the trend of market power in the US since 1980. Two aggregate markups are reported. We can see that the sales- weighted one keeps increasing since 1980 whereas thecost-weighted one flattens out. Thus, whether there is a rise in the aggregate markup since late 1980s crucially depends on the way we aggregate firm-level markup. Figure 1.11 presents the cost-weighted markup distribution. A similar trend is observed that since the late 1980s, the markup distribution becomes stable.
The model cannot generate the rapid rise ofsales-weighted markup. De Loecker et al. (2018) decompose the rise of sales-weighted markup and find that the rise of sales-weighted markup is driven by reallocation component since late 1980s.50 This observation poses a challenge on using the models with variable
markups to explain the rise of markup since reallocation of sales generally indicates a reallocation of production cost as well. Thus, sales-weighted and
cost-weighted markup tend to move in the same direction.
1.7.2
Marketing and economic development
Marketing spending is correlated with the level of economic development. Fig- ure 1.12 plots the media advertising intensity against GDP per capita for multiple countries from 1981 to 2011. Two observations can be made from the figure. Firstly, advertising intensity is positively correlated with income level across countries, and for a given country, it increases along the path of eco- nomic development. Secondly, there is a considerable variation in the income level for a given level of advertising intensity.
50Using Census data in the US, Autor et al. (2017) document the rise of aggregate markup
Figure 1.10: Aggregate markup trend in Compustat data
Figure 1.11: Markup dispersion in Compustat data
Table 1.9: Aggregate markup and markup distribution
Increase in β 0 ↑10% ↑20% ↑30%
cost-weighted markup distribution
aggregate markup 1.1522 1.1525 1.1528 1.1530 p25 markup 1.0856 1.0857 1.0858 1.0860 p50 markup 1.1462 1.1465 1.1468 1.1471 p75 markup 1.1939 1.1941 1.1943 1.1946 p90 markup 1.2429 1.2434 1.2439 1.2445
sales-weighted markup distribution
Figure 1.12: Advertising and economic development
Source: Figure 5 in Corrado and Hao (2014b).
In the model, aggregate marketing intensity is given by W La P C = P kβ(1−uk)Xk W Lkp W Lp µ ,
which depends on four elements. The first one is the returns to scale of market- ing β. Along the process of economic development and technology adoption, new channels and platforms for marketing emerge, which increase the returns to marketing activities. The second component is production-cost share of each industry W Lkp
W Lp . Since appeal tends to be more important for service industries,
structural transformation that reallocates resources towards services would increase the aggregate marketing intensity. Thirdly, aggregate marketing in- tensity is negatively correlated with aggregate markup in the economy. Lastly, the country with more intense signaling competition, i.e., higher Xk, tends to
spend more on marketing. Therefore, given the returns to marketing activi- ties, a high-income country that produces high-quality products could have a relatively low level of marketing intensity due to a higher level of markup or lack of competition among firms.
1.7.3
Marketing vs Advertising
In the firm-level data, I use advertising expenses to measure the marketing ex- pense of the firm. Given there is a shift of marketing practices fromOutbound
creasing since 2000. In this section, I show that using a fraction of marketing spending to measure MPCR would generate exactly the same cross-sectional elasticity pattern with markup. Thus, this measurement issue will not affect the results on the MPCR-markup elasticity and its co-movement with aggre- gate marketing intensity.
Suppose appeal is produced by combining two elements of marketing input using a Cobb-Douglas production function: Φ = QMβ where M =Lχ
a1L 1−χ a2 .51
Then the cost function of appeal is given by C(Φ) = χχ(1−χW)1−χ
Φ
Q
β1
=
W La1 +W La2 = W La. The presence of two components within marketing
function implies a level shift in the cost function of Φ. If we re-do the algebra in the modeling section, this shift has no impact on the distribution of firm size q and the extent of information distortion X. Intuitively, the level shift of the cost function of appeal will not change either the dispersion of sales across firms or the incentive to mimic other firms. Even if we use a fraction of marketing spending to measure MPCR, the MPCR-markup elasticity depends only on the information distortion X, which is determined by the aggregate returns of marketing technologyβ. Thus, measuring MPCR using advertising expense would result in an identical level of MPCR-markup elasticity
W La1(Qn)
W Lp(Qn)
= χβ(1−u)
α X(Qn).
Similarly, for the co-movement pattern, the change of χ within marketing technology is irrelevant for both the MPCR-markup elasticity and aggregate marketing intensity.