Globalization and Skyscrapers: The Role of Architects in Latin America and Southeast Asia

Abstract

This paper performs a theoretical-empirical analysis of the effect of globalization on buildings’ height in Latin America and Southeast Asia. We develop a principal-agent model where global architecture firms use extra height as advertising in the tallest building per city, adding it to what determined by cities’ fundamentals of economic and geographic size. The model develops two ideas: (1) global architects have prestige advantages that allow them to add extra height, an advertising feature of their technical expertise and specialist competence, and (2) international architects of the Global South have additional reasons to use extra height compared to their Global North peers. The model is tested using a 2000–2018 panel database comprising 55 cities (from 25 countries). We find that in addition to economic and geography fundamentals, globalization and the location of architecture firms are strong determinants of buildings’ height, as predicted in our theoretical model of height as advertising.

HIGHLIGHTS

1. We develop a behavioral high-rise construction model for Global South cities.

2. The model is tested using panel data on 55 cities from 25 countries.

3. Buildings’ height is determined by city fundamentals and globalization.

4. Regression results prove the effect of architects on extra height.

Keywords:

• Global cities
• high-rise construction
• Latin America
• Southeast Asia
• principal-agent model

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Land availability refers to the total extension of the urbanized area for two reasons: (1) In theory, all the urban land is in the market at every point in time, not just fringe or vacant land. Furthermore, (2) highrise buildings are usually redevelopment projects in already consolidated central or pericentral locations.

2 The figures correct the endogeneity between the variables: The larger urban economies are, by definition, the geographically larger metropolitan areas.

3 Fuerst et al. (2009) perform a similar exercise with USA data but emphasize differences in price, not in height. Nase et al. (2019), emphasize the demand side (hedonic) contributions to building height in Amsterdam.

4 A more general discussion of principal-agent models is offered in Sappington (1991).

5 Sappington (1991, p. 45) highlights the essence of the principal-agent problem: “If you [the principal] want something done right, do it yourself.” However, in our complex economy, principals cannot do everything by themselves and, thus, need agents.

6 For example, we could have revenues linearly positively related to extra height $E,$ like the $R^D+\beta E$ with $0<\beta$ (instead of $\beta =0,$ as in Equation 2(2) $${\mathrm{\Pi }}^D=R^D-(w+\gamma (H^A-H^D\left)\right)$$(2) ). It can be verified that, by doing so, the predicted $0<E^{*}$ would not change.

7 This is a fairly reasonable range of potential $\gamma$ values. For instance, Barr (2010) estimated the range of the cost parameter for skyscraper construction in New York to be between 0.630 and 0.430, which is well behind 2.

8 We are explicitly saying U^D,ot denotes “utility,” not “profits,” in order to highlight $D$ preference for skyscraper development may be motivated by subjective (utility) reasons, more than by purely profit-maximizing ones.

9 This does not necessarily mean that the interactions are actually occurring in different periods. Time in this type of models is “logical,” not chronological.

10 To obtain that $t\in (0,1)$ we assume that $E<a.$ But it is worth mentioning that we could have assumed, somewhat differently, that $t=E-a$ (which also makes $t$ increasing in $E$ and decreasing in $a$). For having $t\in (0,1),$ the latter implies assuming that $a<E<1+a.$ And, under this assumption, the key finding that $0<E^{*}$ would remain.

11 The reader can verify that the second-order condition for a maximum is also satisfied.

12 If the architect rate for participating in a project is the same in both the Global North and South, the lower financial value of Global South projects causes them to have a lower income anyway.

13 Determined by city fundamentals and independent from $E.$

14 That Equation (6)(6) $$\underset{w}{\min :}\mathrm{ }w\left(1-\frac{\gamma }{2}\right)+\frac{\gamma a}{2(1-b)}.$$(6) is linearly increasing in $w$ guarantees that $w^{*}=0$ is indeed a minimum.

15 It is worth highlighting that this result does not hinge on the assumption that $A$ fallback position $(b)$ is positive, as we could assume that $b=0$ and, nevertheless, obtain that ${0<\frac{a}{2}=E}^{*}.$ In fact, the ${0<E}^{*}$ result hinges on the assumption that $A$ command prestige: $0<a.$

16 We must clarify that extra height is not the only way in which architects seek international recognition. Truly renowned global architects do not need extra height as advertisement, for example, the Sydney Opera House or the New York Guggenheim Museum. However, our model focuses on the less influential, international architects of the Global South. For them, adding extra height is one of their few tools to achieve global notoriety.

17 The Southeast Asian sample includes two high-income countries: Singapore and Taiwan. It also includes Bangladesh, a country not usually included as Southeast Asia. We include these countries to increase the sample size.

18 This would be the preferred (optimal) height for the developer in our principal-agent model. It is the baseline without extra height as advertising.

19 Eight cities in our database did not have any building at more than 100 meters in 2000: Cartagena (Colombia), Puebla (Mexico), Rosario (Argentina), Santo Domingo (Dominican Republic), Bandar Seri Begawan (Brunei), Phnom Penh (Cambodia), Bandung (Indonesia), and Da Nang (Vietnam). They are included because they reach such a height early in the decade (before 2005). Our panel database is slightly unbalanced $(873/1045 = 83.5\%).$ Such a problem does not have an effect on estimation quality (goodness-of-fit, assumptions tests, interspecifications consistency, and cross-section city effects), as we will see later.

20 Appendix Table A2 performs an alternative exercise, where the dependent variable is the percentage difference between a top building and the second tallest in every city/year. The results still support our theory and reject H-S (2008), because globalization, foreign and foreign-and-regional are positively associated to the heights, while the regressions have low goodness-of-fit, the residuals are not normal and do not have consistent panel effects. This alternative exercise cannot be theoretically associated to fundamentals because they explain the height in levels, not their percentage difference. We are using the same specifications of Table 3 to maintain consistency; however, the lack of theoretical soundness of these models is evident in their lack of robustness.

21 This particularity has reverted in some high-income countries’ cities (Marfella, 2016), which also seems to be changing in our case study: Faster height growth in residential than in office buildings.