1
It Was the Economy, Stupid
Stephen A. Kowalewski
Were the major cycles of growth and decline in Mesoamerican civilization (and other urban societies) caused by uncommon (in alphabetical order) aliens, droughts, eruptions, exhaustions, invaders, or raptures? This chapter reviews a theory of the ancient Mesoamerican economy, tests some of its expectations, and proposes that common economic forces would be a reasonable and sufficient cause for episodes of growth and decline. Despite problems of archaeological specification, there is sufficient reason to develop this line of research, in which preindustrial urban societies are treated as subject not to exotic forces but to things familiar to our own experience. Our field is weak in general theory concerning the long-term dynamics of urban societies. Explanations tend to be ad hoc, particularistic, long-since discredited, or reliant on exogenous causes; social science needs to identify regularities and processes in its domain, society itself—and central to social life is economics. Hence the title of this chapter: in 1992 the political advisor James Carville kept admonishing the Bill Clinton campaign to stick to the main issue, insisting it’s “the economy, stupid” (Kelly 1992).
In a recent essay (Kowalewski 2012), I explored in conceptual or theoretical terms how the ancient Mesoamerican economy worked. Here I develop observable implications of the theory, using data from five decades of regional archaeological survey in highland Oaxaca.
Theoretical Base
What follows is a general model of the ancient Mesoamerican economy. As a model it is not an empirical description. To distinguish the two, I use present tense for the model and past tense for the past observable world. The theoretical model is general, not designed for one place or time in Mesoamerican history. It is formal in the sense of using general concepts or principles not tied to specific cultural contexts. And in its economic anthropology style it is formalist rather than substantivist, as explained by Cook (1966).
Formal models such as what follows are not made from sensory data, but from principles; they are not data but they tell you what data are. The purpose of models is to run them up against real situations so one can see whether something was acting as if it were doing so according to one model’s principles or those of another. The model is an ideal type designed for comparison—data against model, and by means of data, model against model (e.g., Apostel 1961; Braudel 1970; Clarke 1972; Weber 1947).
As I present parts of the theory here, I include a few key references to the relevant Mesoamerican archaeology and history. In themselves these sources do not “prove” the principle or premise, but only show that applying the premise here is reasonable and that it can have connection to Mesoamerica.
The following paragraphs describe this theoretical model of the ancient Mesoamerican economy. This economy works by market principles of supply and demand. Many scholars would deny that it is a market economy. Perhaps calling it a commercial economy (commerce means goods exchange) would be more agreeable. The real task is to explain how the economy functions and how economy and society shape each other.
I begin with the actors, which are households, mostly but not entirely smallholders. Households do not produce all the goods and services they consume. They desire to consume and they produce for exchange. Goods and services (including labor) circulate. The household is the firm, the marketer, and the consumer all in one. These premises are realistic given the abundant archaeological studies showing that in Mesoamerica the household or house was the unit of production and consumption (e.g., Hendon 1996; Hirth 2009, 2013; Robin 2003a; Santley and Hirth 1993). (That the household is the firm and the locus of specialization, and that there are so many household-firms, suggests that the ancient economy is a better case of the economist’s “perfect competition” than is oligopoly capitalism.)
Products in ancient Mesoamerica are elaborated, specialized, differentiated, subject to fashion, and consumed in great quantities. Notice that the properties just listed are not exclusive to industrial manufacturing. This blurs the distinction in Western economics between agricultural and industrial sectors and in anthropology between industrial and preindustrial economies. Again, these premises about products are realistic (see Berdan, chapter 6, this volume), and for examples from the large literature on Mesoamerican technologies, Sahagún (1950–1982) for sixteenth-century historical descriptions, Feinman and Nicholas (2000) on shell, Healan (2011) on obsidian, and Tarkanian and Hosler (2011) on rubber.
Households produce for the market and consume from the market. Urban concentration increases exchange and market dependence. Given Mesoamerica’s large urban and rural populations, the demand for goods and services requires widespread and daily participation in exchange. There is a high volume of transactions. In fact, urbanization rates in Mesoamerica were comparable in many respects to those in preindustrial Mesopotamia, China, and Europe (Kowalewski 1990). On theoretical grounds, Kohler et al. (2000) argued for anonymous market exchange rather than personal reciprocal exchange for late prehistoric pueblos in New Mexico, where populations and population density were much lower than in Mesoamerica.
Exchange takes place by means of goods markets (in and outside of marketplaces) in which prices are determined by supply and demand (a key characteristic of market economy, e.g., Pryor 1977). Mesoamerica’s systems of periodic markets are well known. Comparative research by Richard Blanton (1985) showed that the density of market places was equal to or higher than that in preindustrial Europe and China. Recent archaeological studies of markets and marketing in Mesoamerica include for example LeCount (chapter 7, this volume) and Stark and Garraty (2010). Some prices may be set by tradition or local law but these too must vary in response to supply and demand in the longer run. Exchange also takes place through nonmarket mechanisms. The relative importance and relationship between market and nonmarket mechanisms is an important factor in the dynamics of the economy, as discussed below.
Given the high volumes of exchange, it is improbable that transactions take place without credit and culturally defined media of exchange. This is not barter. (Graeber 2011:21–41 argues that the barter economy is a myth.) There are multiple commodity monies, used as means of exchange, standards of value, stored value, and means of credit or account. For Mesoamerican monies, Millon (1955) compiled historic descriptions of cacao as money (see also Smith 2012).
The degree to which Mesoamerica diverged from the Old World in the matter of money has not been seriously investigated. Mesoamerica did not have state-minted coins nor carefully measured bullion. Nor is there evidence of the strong silver-staple grain nexus seen in the Near East (e.g., Davies 2002; Powell 1996; Wray 2004). Whether Mesoamerican commodity monies were a more popular, less statist method of exchange is an interesting speculative problem best left aside for now.
The preponderance of transactions takes place outside the tax or tribute system, that is, among the masses of household producer-marketer-consumer firms. The economy is largely structured by the principles governing the household consumption-exchange-production circuit, not tribute. On tribute in Mesoamerica, see for example Berdan and Anawalt (1992) for the Aztec Matrícula de Tributos/Codex Mendoza and Landa (1941) for Yucatán. The impressive total volume of the Aztec tribute needs to be tempered with per capita measurements (e.g., Kowalewski et al. 2010). The weight of the tribute burden would have varied with state power, core versus periphery position, local resources, and other factors.
The degree of regional specialization and division of labor is a function of transaction costs. This premise follows models developed by Krugman and colleagues (Fujita et al. 1999; Krugman 1980). Although these models explicitly privilege industry over agriculture, I see no reason to assume that the process only works with industry, since Mesoamerican agriculture and forestry could themselves be quite differentiated and dynamic.
The production, consumption, and exchange activities of firms—households, that is—drives the economy. In turn the aggregate household behavior sets the conditions that actors have to deal with in their affairs. The economy works by the invisible hand, or the aggregate effect of households consuming, exchanging, and producing. Lords and the wealthy, operating as large houses, manipulate and take advantage of exchange and accumulated labor, but this is not a state or command economy.
Consumption, production, savings, and investment are variable, not constant. Likewise, exchange through market versus nonmarket institutions is variable, not constant. A key factor influencing household behavior in these things is access to efficient markets. If access is poor or markets cannot deliver goods at acceptable prices, households can withdraw from participation; if access and efficiency are better, that encourages participation. High levels of demand encourage more market participation, all other things being equal, and withdrawal from the market, if sufficiently prevalent, makes the market less efficient. Participation depends on prices, expectations, demand, trust, and confidence. Here we have the ingredients for volatility, for good times and bad, and cycles of boom and bust.
These ideas were more developed in the longer article (Kowalewski 2012). In this essay, I ground aspects of the larger theory to archaeological data from highland Oaxaca, especially several features that should be manifest at the regional scale. Testing a broad theory of economy has to be done piece by piece, some of the pieces are more amenable than others; features that are better addressed at macro, local, or household scales are beyond the scope of this chapter.
If the ancient economy operated by market principles, then certain expectations follow. The distribution of cities should conform more to commercial than political needs. Regional specializations should take hold in response to market mechanisms. Economic cycles of growth and decline should affect rates of production and consumption; that is, output and consumption should be variable, not fixed, and they should be related to market integration. The distribution of material wealth among households should be strongly influenced by market participation. Some of these expectations can be tested fairly easily but others are more difficult because of the magnitude of the data requirements.
City Systems
To what degree was the distribution of cities determined by commercial factors—the invisible hand? Alternatively, did the landscape reflect the vision of the kingdom—the visible hand of power? We can assess whether the invisible hand or the naked hand was the stronger, because the two processes lead to distinctive settlement patterns.
In preview, the settlement pattern difference is this: if the hand of power is stronger, exchange is oriented toward a single center in an exclusive territory. If the invisible hand is stronger, exchange is distributed among multiple nodes in a network, boundaries are permeable, and commerce draws participants close together regardless of political affiliation.
Here I develop and use a simple model to measure where regional city systems fit on a continuum between these two polar positions, the political and the commercial. In the former, rulers place their capitals at a maximum distance from one another in order to have exclusive sovereign control over as much territory and as many subjects as possible, with a buffer zone between themselves and their counterparts. Christaller thought of this as a sociopolitical, noneconomic principle, which he called separation: “The ideal . . . has the nucleus as the capital (a central place of a higher rank), around it, a wreath of satellites places of lesser importance, and toward the edge of the region a thinning population density—and even uninhabited areas” (Christaller 1966:77).
Christaller’s separation principle resembles closely an idealized, isolated Mesoamerican city-state, in Aztec terms, the altepetl. Mesoamerica had many cities, states tended to be small although they could be combined into larger alliances or empires, the state had a capital that was the largest city in its domain, and the city-state had a longstanding territorial nucleus. In Mesoamerica, the city did not have the legal autonomy that many cities in Europe did, and the state was not always defined as a contiguous territory, since hereditary rulers sometimes had subjects or holdings in scattered places (Hirth 2003; Smith 2008). These particularities aside, the altepetl was a political, autonomous entity, defined by itself without relation to its neighbors, and it resided in a persistent core territory. The separation principle is thus a good representation of the Mesoamerican political vision.
I needed another ideal model for the other, commercial end of the continuum. I considered Christaller’s supply or marketing principle: “The system of central places has been developed, on the basis of the range of the central goods, from the point of view that all parts of the region are supplied with all conceivable central goods from the minimum possible number of functioning central places” (Christaller 1966:72). This principle entails four assumptions that were difficult for me to make: that I knew the central goods (estimated population size of centers is what we have), goods were supplied everywhere, they were supplied by a minimal number of centers (Christaller wanted efficiency), and land was an isotropic plain (highland Oaxaca is not). I needed a simpler model.
I found the basic idea for a simpler model in Christaller’s results and conclusions. He had observed that the wealthier, more populous regions in southern Germany had more numerous high-order central places, which tended to cluster and be close to one another, but the poorer regions had fewer high-order places and other central places were more widely spaced. For Germany as a whole, cities also tended to cluster together in the wealthy regions (such as the Rhine-Ruhr Valleys), whereas poorer regions had fewer, more separated, but larger high-order centers (Berlin and Munich, for example) (Christaller 1966:193; Smith and Branom 1937). This contrast gave me the direction I needed.
When commercial activity dominates, cities tend to cluster, as buyers and sellers of central-place goods try to increase the number of exchange partners within their reach. Cities are contagious, they are attracted to one another. But they are not totally drawn into one megacenter, because neither economic nor political power is monopolized and because competing centers serve and draw from hinterland customers and producers. Unlike the conceptual model of the independent kingdom, cities created in the commercial world are expressly situated in relation to each other in a wider world. The result is a galaxy or cluster of roughly equal-size centers located near one another, with overlapping wedge- or pie-shaped hinterlands expanding outward. In this manner the entire developed region is served not by one monopolistic center but by competing centers that may offer varying ranges of specialized goods. The borders of political territories are permeable and population does not thin out toward borders. In essence, the commercial economy develops multiple high-order central places relatively close to one another, without regard to borders.
City distributions can be assessed against the contrasting expectations of the ideal separation (altepetl) model versus this ideal commercial model. Richard Blanton (1996) mapped the distribution of Aztec altepetl centers overlaid on a reconstructed understanding of the regional market system in the Basin of Mexico. He was able to show that the market system was already an influence on the location of centers in Early Aztec times, and that the market network filled in and became more integrated in the Late Aztec period.
In the next several pages, I evaluate the spatial distribution of central places using three slightly different approaches: at a macroregional scale using only the largest cities; a regional scale that adds middle-size central places; and regional-scale analysis of large and middle-size central places that examines the location of centers with regard to the borders of political territories. Factors such as land quality, topography, and transport differentials influence city distributions. I can control these other factors partially by comparing change over several periods of time in the same place.
Figure 1.1 shows the regional archaeological survey coverage, now over 7,500 km2, from the Mixteca Alta to the Valley of Oaxaca. The data are from the following sources: Tamazulapan/Tejupan (Byland 1980); Coixtlahuaca (García Ayala 2011) and work by this author’s project in progress; Cuicatlán (Spencer and Redmond 1997); Central Mixteca Alta (Kowalewski et al. 2009; Pérez Rodríguez, Anderson, and Neff 2011; Spores and Robles García 2007); Nochixtlán (Byland and Pohl 1994; Plunket 1983; Pohl and Byland 1990; Spores 1972); Peñoles (Finsten 1996; Smith 1993); Sosola/Tenango (Drennan 1989); Valley of Oaxaca (Blanton 1978; Blanton et al. 1982; Kowalewski et al. 1989); Sola (Balkansky 2002); Ejutla (Feinman and Nicholas 1990); Miahuatlán (Markman 1981); Guirún (Feinman and Nicholas 2004).
Figure 1.1. Spatial distribution and sizes of cities. (Data sources are cited in the text.)
There are limitations on the comparability of data sets gathered over five decades by different projects, but if I restrict the analysis to those things that can be compared, especially the largest settlements, these survey data meet present purposes quite well. The older surveys have broader periodizations than the more recent ones. For example the Nochixtlán survey used a general Ramos period (Late and Terminal Formative) whereas the Central Mixteca Alta survey split Ramos into Early and Late. Because of this Nochixtlán may be somewhat underrepresented in the discussion that follows. I am not including the Mixteca Baja, west of the area under consideration (Rivera Guzmán 1999), because the surveys there are not contiguous to those in the Mixteca Alta. For discussion of comparability problems with highland Mesoamerican survey data see Smith (2002).
The first approach concentrates on just the largest cities—that is, all settlements with estimated populations greater than 5,000. I am using this limit of 5,000 inhabitants for all time periods to improve comparability, to exclude sites that were just overgrown villages without the full array of urban central-place functions, and to simplify the visual presentation. Figure 1.1 shows the distributions and relative sizes of the cities as I defined them, the top-ranked central places with populations over 5,000; the maps do not show the many smaller cities and towns. The time periods are the Terminal Formative (Monte Albán II in the Valley of Oaxaca and Late Ramos in the Mixteca Alta), Early Classic (IIIA and Early Las Flores), Late Classic/Epiclassic (IIIB-IV and Late Las Flores), and Postclassic, which is mostly but not exclusively Late Postclassic (V and Natividad).
Multiple urban centers emerged in the Terminal Formative. The principal ones were Monte Albán and Huamelulpan, and the third, Ñiaxugue-Naduza in Coixtlahuaca, has the minimum size but still might be suspect as an urban place. Monte Albán had a well-developed system of towns in its hinterland but the same cannot be said for the Mixteca Alta, where in this time period much of the rural area was abandoned and population concentrated into the immediate environs of the large center. Compared to the later periods, the two (or three) cities were quite distant from one another, as in the ideal type of the political landscape.
Both the Valley of Oaxaca and the Mixteca Alta had developed city systems in the Early Classic. There were clusters of four leading cities in both regions. These largest cities were quite close to one another (20 km in the Valley of Oaxaca and 8–18 km in the Mixteca Alta). The maximum distance between these centers—the size of the cluster along its long axis—was 47 km in the Valley and 23 km in the Mixteca Alta. Both regions had substantial development of smaller cities and towns not shown on the map (internally and also extending to the south in both cases). Using the political/commercial landscape model, I suggest two observations: (1) that within each region commercial forces attracted the largest centers toward each other; and (2) that there was still quite a bit of distancing between the major regions (the Mixteca Alta and the Valley of Oaxaca).
In the Late Classic/Epiclassic, the Mixteca Alta did have some settlement, but many areas seem to have been totally abandoned. The Valley of Oaxaca had Monte Albán, which grew to its maximum and then collapsed in this period, two places that seem to have grown early and then collapsed, and two that seem to have grown larger as the others fell. All five of these disappeared as cities by the end of the Epiclassic.
Finally, the Postclassic had a much larger urban system. This was the time of maximum population in all regions. There were two large clusters of top-ranked cities, and all regions had strong development of lesser cities and towns. The Valley of Oaxaca had a more dispersed settlement pattern, while in the Mixteca Alta a greater proportion resided in large cities. The second-largest city in the Mixteca Alta is Cerro Jazmín, which according to the most recent study had most of its occupation a bit earlier in the Postclassic (Pérez Rodríguez, Anderson, and Neff 2011). The other very large city was Inguiteria or Coixtlahuaca, known to be quite late.
In the Postclassic, cities clustered together. One clump in the Mixteca Alta measures 77 km in maximum length, the other in the Valley of Oaxaca extends over 49 km, suggesting attraction and the commercial side of the continuum. Within each of these clusters, spacing was rather close—the average nearest-neighbor distance between these top-ranked cities was about 10 km, a two-way trip easily walked in a day. Perhaps the very large sizes of Jazmín and Inguiteria indicate another level emerging at the top of the urban hierarchy.
The Postclassic settlement patterns were not simply a result of population growth and filling-in of available space. Instead, people aggregated into cities that were distributed in a particular way. Within regional systems the largest cities were closely spaced. The mutual attraction of the top cities, that is, the persistent clustering into hierarchically organized, regional systems of cities, suggests commercial forces operating at the regional and intraregional scales.
Moving to the second approach, I expand the range of central places to include middle-size places. Three decades ago Jill Appel (1986) used Christaller’s central-place models to study Valley of Oaxaca central-place patterns. She tested whether the Early Classic and Late Postclassic conformed more to a k = 3 pattern (each center equally spaced from three higher-level centers, ideally) or a k = 4 pattern (each center equally spaced between two higher ranking places). A k = 3 system is most efficient for rural consumers where transport is not developed; k = 4 optimizes the movement of bulk goods along straight roads between centers. Appel found that the settlement pattern fit the k = 3, rural-retail model, especially the Late Postclassic but also the Early Classic.
The Mixteca Alta has many small valleys surrounded by mountains, quite different from the open plain of the Valley of Oaxaca. I use our Central Mixteca Alta survey area to examine central place distributions. The size of this study area is 1622 km2, roughly 44 × 37 km. The top-ranked population centers in the Classic and Postclassic had about 17,000 and 32,000 inhabitants, respectively, and considering the top 30 centers, the smallest had about 1,000 inhabitants (table 1.1).
Table 1.1. Spacing and size range of centers in the central Mixteca Alta (NN = nearest neighbor)
Classic | Postclassic | 1990 | |
---|---|---|---|
Top Twenty Centers | |||
Pop. of Largest Center | 17,180 | 31,996 | 9,555 |
Pop. of Smallest Center | 1305 | 2145 | 437 |
Average NN Distance (km) | 5.0 | 4.0 | 5.9 |
NN Statistic R | 1.1 | 0.89 | 1.32 |
Top Thirty Centers | |||
---|---|---|---|
Pop. of Largest Center | 17,180 | 31,996 | |
Pop. of Smallest Center | 963 | 1475 | |
Average NN Distance (km) | 3.6 | 3.3 | |
NN Statistic R | 0.80 | 0.74 |
Source: Kowalewski et al. 2009.
Half of the total population of the area resided in the top 20 centers (in both time periods), which is a high rate of urbanization. The top population centers were physically close to one another—the average distances from each center to its nearest neighbor center were only 3.3–5.0 km (depending on time period), less than an hour’s walking time. This aggregation into central places situated close to one another is consistent with the model of commercial exchange.
The spatial distribution of population centers is expected to be even or regular under the separation principle but more clustered under the commercial principle. To assess this expectation, I use the nearest-neighbor statistic R (the simplest uncorrected measure in Pinder et al. 1979). The conventional interpretation of R is that 0 is perfectly clustered, 1.0 is random (random distributions have some clustering), and 2.15 is perfectly even or regular. Since R is quite sensitive to the size of the study area, and since the Mixteca Alta’s natural resources are clumped and not evenly distributed, a strict interpretation of the numeric values is not as informative as their relative tendencies. For the top 20 population centers, the nearest neighbor R values are 1.10 for the Early Classic, 0.89 for the Postclassic, and 1.32 for 1990; for the top 30 centers, R = 0.80 for the Classic and 0.74 for the Postclassic (table 1.1). (I did not calculate, R for the top 30 centers in 1990 because that list would extend to small places with no central functions). In the prehispanic periods, centers tended toward clustering, with R dipping below 1.0 (random) in three of the four samples. Postclassic cities were more clustered than Early Classic cities. The recent settlements are more evenly distributed, as expected since most are the administrative centers for their municipalities. My interpretation is that the nearest-neighbor statistic is discriminating between the dominant administrative principle of recent times versus the more commercial principle of prehispanic times.
My third analytical approach to the spatial distribution of central places considers political territories and their borders. Our work in the Central Mixteca Alta (Kowalewski et al. 2009) used both historical and archaeological information to identify and describe the development of the ñuu, the small kingdoms that could be independent states or could be combined with other ñuu into a larger state (yuhuitayu). The ñuu is like the altepetl. If human settlement were dictated by administrative needs, there would be civic-ceremonial buildings at a capital, the capital should be central to the territorial core of the polity, and the demographic center of gravity should also be in the territorial center. Capitals of different polities should be as far from one another as the degree of packing of polities permits (the separation principle). Thus the relevant variables are the specific locations of the civic-ceremonial centers (the palaces), the borders and geographical centers of these irregularly shaped territories, and the locations of the major populated places. Not counting the polities that we only partially surveyed because they extended beyond our study area to the east or to the west, we identified 17 ñuu in the Central Mixteca Alta.
In the Late Postclassic, the palaces were usually at the demographic center of the ñuu (e.g., Teposcolula), in the largest settlement (Magdalena/San Isidro Peñasco), or at another place more demographically central (Achiutla) (figure 1.2). This relationship conforms to the political model.
Figure 1.2. Postclassic sites in the Central Mixteca Alta. (Map by John F. Chamblee and John C. Burns.)
But there was also a strong tendency for the leading demographic center not to be situated in the geographical center of the polity. Instead the demographic center was attracted toward similar population concentrations in neighboring polities. For example the “inner basin” group of the 10 ñuu in the southern third of the study area all had populations strongly attracted to their neighbors. Apparently the political center was not in this cluster at all—it was in Tlaxiaco, 15 km to the west. A second cluster is made up of Yucuxaco, Huamelulpan, and Tayata. In the north all the polities have their populations skewed toward population centers in Teposcolula, which was also the seat of a yuhuitayu lord. (Yodobada and Lagunas also have a second attraction toward the Tejupan center, just north of the study area.)
Quantitative tests reinforce these observations. Cities and large towns tended to be pulled toward the borders of the ñuu instead of being located in the polity’s geographic center. The mean distance from geographic center to nearest border is 2.7 km. This represents the maximum potential separation of cities from a neighboring territory. Since I measured from the centers of the settlements, the minimum possible distance to the border is a bit larger than zero, about 0.2 km for the smaller of these centers. How do the observed locations for the largest population centers in each polity fall along this scale? In the Early Classic the mean distance to border was 1.7 km; in the Postclassic it was 1.5 km, and in 1990 it was 2.0 km (p = < 0.05 for the three groups). The largest cities or towns were pulled from the center toward the edge of the territory in prehispanic times, and in the Postclassic they were actually closer to border than to the geographical center. The recent locations conform much more to the administrative expectation than the prehispanic.
That populations would be attracted away from the geographical center of the polity and toward each other is not accounted for by the purely political model of the ñuu or altepetl. Economic—commercial/labor/consumption—forces were stronger than the political ideal. One might wryly propose that the people located as they wished and the kings came to perch on top of them, but lords were also involved in commerce and wanted to be where the action was.
To summarize, if the commercial economy determined settlement pattern more so than political forces, then cities should cluster. The three means of analysis all indicate the same result. At the broad macroregional scale the largest cities did tend to cluster, somewhat in the Early Classic and more so in the Postclassic. Within regions, the top- and middle-ranked central places also tended to cluster, noticeably in the Early Classic and more so in the Postclassic. Large and middle-ranked population centers were situated near each other, regardless of political borders. In the Late and Terminal Formative, political forces were relatively stronger, but by the Early Classic and especially in the Postclassic, settlement patterns were more heavily determined by commercial forces.
Mesoamerican stelae, codices, and lienzos depict the indigenous lord’s concept of the kingdom. Apparently no one commissioned monuments to the invisible hand. But the archaeological evidence indicates that by Classic times commercial forces were powerful determinants of settlement pattern.
Regional Specialization
Nineteenth-century economists understood regional or zonal specialization (Ricardo 1821:ch. VII). In market economies, zonal specialization works by comparative advantage. It originates and is maintained by low transaction costs, which include transportation and other costs such as tariffs, bribes, arrangements, and so on. Fargher (2009) provides a cross-cultural analysis of five premodern or early modern states in which there developed strongly differentiated core zones of intensive agriculture and peripheral zones of mixed extensive agriculture along with other activities such as craft production. Since the marginal returns to labor in the more fertile areas are high with intensification, farmers in these zones specialize in farming, even though they could do other things; likewise, producers in less-fertile zones should turn sooner to other activities besides crops because the returns will be comparatively higher (Fargher 2009).
In the early periods in highland Oaxaca the hinterlands of the first urban centers did not show much evidence of this zonal specialization. Agriculture and other sectors were intensified, but this took place everywhere in the urban hinterland and without the clear emergence of zones of specialization.
Zonal specialization in the Valley of Oaxaca emerged in the Early Classic. Settlement patterns indicate intensive agriculture in the fertile northern arm of the valley. Extensive agriculture, cultivation of xerophytic plants, and intensive chipped-stone production and use took place in the drier southern and eastern valley (Kowalewski et al. 1989; Feinman et al. 2007).
There is less evidence of Early Classic zonal specialization in the Mixteca Alta. Differences in settlement types and locations strongly suggest extensive upland farming and intensive farming using terraces, but the zonal patterning does not appear to have been as pronounced as in the Valley of Oaxaca (Kowalewski et al. 2009).
The growth and decline during the Late Classic and Epiclassic created short-lived and fairly small zones of specialization. This was the first time the mountains west of the Valley of Oaxaca were intensively occupied (Finsten 1996; Garvin 1994).(There had been similar high-elevation settlement farther west, in the Central Mixteca Alta, earlier.) It is likely that charcoal, fuel-wood, construction timber, flowers, fruit, other mountain products, and perhaps labor were being drawn into valley markets. Some places in the Valley of Oaxaca, such as Lambityeco, Macuilxochitl, and a number of Etla towns, had remarkable spurts of growth and construction for a few generations, as did several new subdivisions at Monte Albán (Blanton 1978; Lind and Urcid 2010). Growth was fueled by intensive canal irrigation in Etla, floodwater farming in Tlacolula, extensive agriculture in uplands, and exploitation of mountain products. At times, hilltop-terrace towns and villages in the drier south and east may have provided surplus labor. These systems of zonal specialization were fairly small in spatial scale and they probably did not last very long.
In the Late Postclassic zonal specialization is evident in all regions. In the Valley of Oaxaca the northern Etla arm was a zone of intensive agriculture, with much less specialization in other crafts. The eastern (Tlacolula) and southeastern (Ocotlán) zones of the Valley have abundant evidence for craft specialization in chipped and ground stone, pottery, salt, lime, and xerophytic plants. Production in the mountains west of the Valley was at an all-time high.
Interestingly the Valley of Oaxaca regained a strong zonal specialization by the nineteenth century. The zonal pattern mirrors that of the Postclassic: Etla had intensive agriculture and dairy oriented to the urban market; Tlacolula and Ocotlán had more craft specialization (Kowalewski 1995).
In the Mixteca Alta, especially in the Postclassic, labor-and-land intensive agriculture was carried out in core areas, especially on the fertile, loose soils of the Yanhuitlán-Jaltepec geological formations. This specialization involved making and maintaining cross-drainage and contour terraces. These works were so extensive that they covered and transformed the whole landscape. The highest populations are found in these core agricultural areas.
In several outlying subregions of the Central Mixteca Alta, we found very extensive chipped-stone quarries and workshops. Some of the products were destined to population centers in core areas and some may have been used in the processing of upland products.
Archaeological surveys in the Mixteca Alta have not found much evidence of pottery-making places, compared to the Valley of Oaxaca. Although the surveys were conducted with the same methods, the lack of evidence for pottery-making places in the Mixteca Alta conceivably could be an unknown identification problem. But another possibility is that the Mixteca Alta surveys have concentrated in core areas and that the pottery making took place in marginal zones. The recent pottery-making villages are located (with only one exception) in upland places outside the core areas where intensive agriculture is practiced (figure 1.3). Whether this pattern in the historical era was true of the Postclassic is not known.
Figure 1.3. Twentieth-century pottery-making villages and agricultural core areas in the Mixteca Alta. The villages are San Miguel Adequez, Santa Inés del Río, Buenavista Jaltepec, Santo Domingo Tonaltepec, Vista Hermosa Tonaltepec, Río Blanco Tonaltepec, Magdalena Peñasco, San Juan Mixtepec, Rancho Morelos Tinú, San Antonio Nduaxico, Santa María Cuquila, Atatlahuaca, and San Juan Numí (Spores 2007:108; Warner 1976).
Alliances between petty kingdoms, or blocks of ñuu subject to one ruling house, could have promoted internal exchange. Perhaps a consequence of state or royal house expansion was to lower the costs of transactions within the territory of the state or among its subjects. One example is the yuhuitayu of Teposcolula, which was centered in the agricultural core zones in the Teposcolula and Tamazulapan valleys and had subject ñuu in all the surrounding upland valleys. (The holdings of a royal house were not always contiguous, nor were alliances always between contiguous polities, but on the archaeological time scale the patterns of core and periphery are detectable and persistent.)
Another institutional mechanism for lowering transactions costs and promoting specialization and exchange is ritual obligation (e.g., Wells and Davis-Salazar 2007). At the intravillage scale, Monaghan (1995) showed how much exchange was generated by calendrical and other ritual events. At the regional scale in Michoacán, Castilleja (2011) has been documenting ritual exchanges that link Purépecha towns and villages to one another. A few years ago I attended the major fiesta in San Juan Bautista Coixtlahuaca, Oaxaca, a town of a thousand inhabitants, for which the mayordomo spent the equivalent of $150,000 USD, a figure that does not include the offerings, contributions, and expenses of scores of other families. Zonal specialization is symbolized and reinforced in the ritual cycle of this fiesta by outlying villages supplying considerable labor, animals, food, and materials. The fiestas also entail pilgrimages, offerings, and commitments from participants in Veracruz, Puebla, and Mexico City. Participants say these practices are a continuation of precolumbian ties that Coixtlahuaca had as a major international marketplace.
The role of ritual in the economy is somewhat contradictory because participation is very demanding of time and effort, and it is a drain on savings that could potentially go toward capital investment. Yet ritual obligation involves many nonmarket exchanges that stimulate production and consumption. Ritual obligations act as stimulus because they turn household savings into investment, and they pave the way for nonritual exchange. Plus, nonmarket offerings of goods and services contribute to the success of the fiesta, which as a whole has a major and widespread effect on economic activity.
Market Integration
Since at least Early Classic times, regional urban systems had well-developed systems of market places. Market places need accessible, appropriate spaces. We studied potential market places in the Central Mixteca Alta (Pluckhahn and Kowalewski 2003), where the terrain is quite mountainous. The lack of naturally occurring flat spots means that if people were to have market places in appropriate, accessible locations they had to build the space. Our survey found many of these public spaces or plazas. Of course we cannot be sure that each plaza was used as a market, but we can say that these were likely the best potential or possible market places. For the Classic and Postclassic, we identified potential market plazas at a rate of one for every 2,000–4,000 people, or one for every 40 km2. This would be a high market density even if only half the places functioned as markets (cf. Blanton 1985). By comparison, in the 1960s the Nochixtlán cyclical market area had 20 markets (Warner 1976)—one market for every 5,400 people and one for every 280 km2.
At times market systems in adjacent regions are poorly connected; at other times the connections may be so good that we can consider it a single pool of producers and consumers. Greater integration spreads risk; smaller systems are subject to higher probabilities of supply-demand mismatches. The degree of connectedness was a major variable in the cycles of development in civilizations (see Smith 2002 for Mesoamerica).
Market integration is understood as the degree to which prices are coordinated over space and among different commodities. For present purposes, I focus on the spatial aspect of market integration. Under preindustrial conditions for the integration of markets in two different regions, consumer demand should be continuous in the intervening space, transportation and transaction costs should not be prohibitive, and there should be nodes of exchange such that local demands might be satisfied by competitors from both regions. Market integration would thus be unlikely if the two regions were separated by uninhabited area and if market places were too far apart.
We have appropriate data to assess whether some of the conditions for market integration were met in highland Oaxaca. Consider the Mixteca Alta and the Valley of Oaxaca as two regions. Did the intervening area have the population and market centers sufficient to allow the regular exchange of goods and services and were producers and consumers in the same market? The Valley of Oaxaca and the Mixteca Alta are separated by 30 km of mountains, some of which, the Sosola-Tenango area, was surveyed by Drennan and another area, Peñoles, by Finsten and myself (Drennan 1989; Finsten 1996). Again it is helpful to evaluate the relative potential for integration from one period to the next. Figure 1.4 shows the locations of the largest cities, just as in figure 1.1; the numbers in the Sosola-Tenango and Peñoles regions are the numbers of settlements. The Nochixtlán and Etla Valleys are the sectors of the Mixteca Alta and the Valley of Oaxaca closest to the intervening area.
Figure 1.4. Relative interconnectedness of Mixteca Alta and Valley of Oaxaca regions as gauged by the numbers of sites in the intervening mountains.
The Terminal Formative was the least-likely period for market integration, as the mountains between the Mixteca Alta and the Valley of Oaxaca were very sparsely populated.
The Etla arm of the Valley of Oaxaca had some settlement in the Early Classic, a string of towns down the middle of the valley, and this string may have continued with several villages in Peñoles. But overall the mountains were not much occupied and Etla was less important than the highly developed eastern and southern arms of the Valley of Oaxaca and parts of the Mixteca Alta. Substantial growth and development occurred in the two regions, but the intervening area did not play much of a role.
The Late Classic/Epiclassic panel in figure 1.4 shows that in the Mixteca Alta no large cities are identified; except for a few places in the Nochixtlán Valley much of the region appears to have been little occupied. In the Valley of Oaxaca there was strong but stuttering growth in the Late Classic before collapse at the end of the Epiclassic. Monte Albán grew larger than ever and then collapsed. Tlacochahuaya and towns in Etla thrived early and then collapsed; Jalieza and Lambityeco were at their largest a bit later, then declined. The mountains outside the Valley to the north and west had villages and a few towns—some earlier, some later, some a bit of both.
The Late Postclassic had strong urban and rural growth in both the Mixteca Alta and the Valley of Oaxaca, and the strongest development ever in the intervening mountains. Sosola-Tenango and especially Peñoles had multiple towns and villages with plazas. In Peñoles the market orientation over most of the area was toward the Valley of Oaxaca, except for the northwestern quarter, which was oriented west to the Nochixtlán Valley. That most of Peñoles traded more with the Valley of Oaxaca is interesting because historically it is Mixtec-speaking, like its neighbors to the west, and its prehispanic ritual artifacts are mostly in the Mixteca Alta style. Yet the Valley of Oaxaca was predominately Zapotec speaking. Economic ties thus criss-crossed linguistic and cultural boundaries.
In sum the Late Postclassic had the most intensive and most extensive market integration, in which the Valley of Oaxaca and the Mixteca Alta, more than any previous time, were connected together into a single system.
It is more difficult to assess market integration in the sense of equivalent prices for goods and thus equivalent cost for the factors of their production. Blanton and colleagues (Blanton et al. 2005) and Williams (2004), have proposed that certain strategic or world-system goods marked broadscale exchange of common commodities and shaped economic relations. These goods provide another line of evidence for market integration. Obsidian is one such commodity. It was much more common in the two periods of greater spatial integration, the Early Classic and (more so) the Late Postclassic; it is less common in the Terminal Formative and the fragmenting Late Classic/Epiclassic, even in the regions experiencing some growth during that time.
And then there is pottery. I think that in market disintegration, standard bulk pottery and its styles are of poor quality and do not travel very far. This stands to reason and one need look no farther than the G-35 type for a bulk good that perhaps for very good reason did not travel very far, whereas the Postclassic G-3M and polychrome did. Likewise the predominant and tell-tale orange look of Early Classic pottery was followed by multiple producers and the style was widely known in common goods.
But pottery and styles traveling far is scarcely a late thing; otherwise, cross-dating would not work as well as it does in all phases from the Early Formative on. We need to be cautious here. Perhaps with urbanization market integration could have been a factor in pottery quality, but it is not the only relevant variable and there is not a very good understanding of why Formative pots and styles traveled as far and as regularly as they did.
Wealth Stratification
The economy shaped the distribution of wealth and livelihood. Mesoamerican social organization is often described in terms of its class system of nobles, a small class of merchants and luxury crafters with intermediate privileges, commoners landed and landless, and slaves. These are ideal or legal categories; these categories may not correspond to the distribution of wealth (Chase and Chase 1992). An indication of the importance of the market would be the degree of divergence between legal class categories and the distribution of household wealth. Feinman and colleagues studied this problem in detail with well-controlled, excavated houses at El Palmillo, a Classic-period town in the Valley of Oaxaca (Feinman et al. 2007). They observed the distribution of 21 different everyday, uncommon, and ritual objects and found that there were differences in household wealth, but that the differences were continuous and quantitative. Our review (Steere and Kowalewski 2012) of house and artifactual data from excavations and mapping projects in Central Mexico, Oaxaca, and the Maya area concludes that wealth distributions were indeed variable, they tended to be continuous, and they often had a stronger middle than predicted under the nobility/commoner model of social stratification.
Variable Consumption
My understanding of the ancient Mesoamerican economy implies that household consumption was not fixed but variable, and that it varied with market dependence and participation. In 1970 I worked for John Paddock at Lambityeco, a town in the Valley of Oaxaca dating to around AD 700. Paddock talked about Lambityeco as decadent, usually pronouncing the word “decay-dent” (Paddock 1970), because the ancient Lambityecans, to his mind, made almost nothing of artistic quality even though they consumed enormous quantities of stuff. For a relatively short occupation—most of it in just a few generations—Lambityeco certainly did leave a lot of débris. Ancient Mesoamericans did that at some times and places, but not always. The Lambityeco bubble was not only relatively brief, it was also localized.
Measuring household consumption rates is not easy because it requires good chronological control, a way of getting at consumption per unit time, and large, representative samples. It requires quantitative measurement. This is why early professional economists in Western countries campaigned for better statistics (i.e., reliable numbers on such things as manufacturing output, sales, employment, and the flow of money). Understanding a modern economy requires quantitative measurement with reliable data. Why should we think the ancient Mesoamerican economy was any different?
Economic Cycles
In our experience today we understand that poorly controlled markets fluctuate between exuberance, stability, and depression. This is not a new revelation. In the fourteenth century Ibn Khaldûn described economic cycles of growth and decline that lasted about 40 years. He suggested a sophisticated explanation based on movements in wages, prices, supply, demand, and expenditures in different sectors (Boulakia 1971; Khaldûn 1967:2:291ff.; Soofi 1995). Cycles of various periodicities are recognized by modern historians and there is considerable literature on their causes (reviewed by Berry 1991).
Did Mesoamerica have a business cycle? If so, was it faster or slower than in economies with metal coinage? Were there economic waves of longer periods? Does the cyclical notion of time characteristic of some civilizations have to do with people trying to understand and control the good and bad times of economic cycles?
Civilizations have longer-period cycles of growth and decline, unification and disintegration. Prolonged stability or growth is not known. All civilizations cycle. Older ideas about rise and fall (Spengler 1926–1928) were discredited long ago, but they keep returning (as in drought causing the Classic Maya collapse, e.g., Haug et al. 2003). Modern social science has not dealt with the problem very well (McAnany and Yoffee 2010). Anthropologists have typically opted to emphasize particular histories instead of general regularities, or they assert that preindustrial societies were a different species from modern societies and not subject to uniformitarian processes. Admittedly this is a large and difficult problem, and I for one do not have the cross-cultural expertise in this area, but our abdication has left the field open for all sorts of even lesser-qualified advocates for exogenous causes and externalities. The professionals leave the field and the amateurs take over.
Discussion
In this chapter I have tried to ground in archaeological evidence six implications of a general theory of a market-dominated Mesoamerican economy: (1) the spatial distribution of cities, (2) regional specialization, (3) market integration, (4) wealth stratification, (5) variable consumption, and (6) economic cycles. These are observable at the regional scale but some are better studied with locality and household data. The present analysis is limited because it is confined to highland Oaxaca. Some of the implications are fairly well understood and supported; others are reasonable but still speculative.
If cities were located according to the political needs of rulers, they should have been spaced far from one another. This was true in the Terminal Formative. But by the Early Classic they were not far from one another, they clustered together. Cities were even more attracted to one another in the Late Postclassic. Urban attraction is a characteristic of commercial economies but it is not anticipated in the idealized political landscape of the Mesoamerican city-state.
Market systems can grow in spatial scale and regional market systems can have different degrees of articulation with each other. In highland Oaxaca, Terminal Formative market systems did not extend beyond their capital city’s hinterland (small for Huamelulpan, larger for Monte Albán). Regional systems were quite integrated in the Early Classic but they did not extend farther than the region in terms of daily, high-volume exchange. In the Late Postclassic, regional systems were highly integrated internally and the two core regions—the Mixteca Alta and the Valley of Oaxaca—had the highest degree of interconnectedness. The Late Classic/Epiclassic was a time of spatially smaller systems, sometimes well integrated internally, but not so integrated at the interregional scale.
In market economies zones of specialized economic activity develop as transactions costs are sufficiently low and producers find comparative advantage in mutual exchange. This happened in Mesoamerica. In Oaxaca, archaeological evidence of regional specialization is strongest in the Late Postclassic and is also found for the Valley of Oaxaca in the Classic period. It was weakly developed in the Early Classic in the Mixteca Alta and there is little indication of it in the Terminal Formative. Because regional or zonal specialization has such an impact on local and household economies, linking variation in domestic production to role in regional system is a promising area for further testing.
Market economies create wealth disparities and continua in wealth distributions that deviate from expectations that wealth would be determined by legal status. Data on household wealth in numerous archaeological cases from different times and places in Mesoamerica do indicate both substantial variation and continuous (as opposed to categorical) variation. Variability in household wealth needs to be linked to other economic factors, such as market integration.
The general theory does not assume a static peasant with fixed need or wants; instead, consumption should be variable, depending in large part on market integration. In principle variable consumption rates are measurable, but practically this seems to me to be a knotty problem with archaeological data.
The commercial economy structured Mesoamerican society in fundamental things: where people lived, how they made a living, their standard of living, the rhythms of life, and so on. By the later periods the distribution of cities, towns, and villages and patterns of regional specializations were consistent with the expectations of market economics. I use the term “market” in a broad sense that does not imply capitalist mechanisms or special institutions derived in the experience of industrialism. Integration of commercial systems was associated with increased production, consumption, market dependence, and demographic growth; market disintegration was associated with decline. I argue that the Mesoamerican economy, like other urbanized economies, underwent such cycles. Still, the intensity and speed of these movements in this commodity-money economy are not yet understood.
Mesoamerica does seem to belong in that set of civilizations within which it is totally appropriate to explore how market economies behave and how they might be better understood than if we relied only on the narrower range of experience of select Western countries in the past couple of centuries. Such a “modernist” stance, in which the past is not a caste apart but a relevant kindred of our own experience, has enlivened studies of the economic history of the Near East. Silver’s summary comment on Polanyi’s view of markets in the Near East applies just as well to Mesoamerica: “It is incorrect to magnify the economic flows of temple or palace to Amazonian proportions while shrinking the market to a mere brook” (Silver 1983:829). Temin (2013) has used Roman data to challenge the claim that the ancient economy was incapable of real growth and had only Malthusian growth. Similarly for Mesoamerica, Stark and colleagues (1998) pointed out nearly 20 years ago that Mesoamerica’s cotton sector exhibited growth and dynamism sufficiently powerful to affect important aspects of society and the regional and interregional economy. Stark has explicitly taken up the question of economic growth, describing difficulties in archaeological measures, but concluding nevertheless that in precolumbian Mesoamerica “economic growth is largely episodic or unstable, but nevertheless exhibits some cumulative effects” (Stark 2013).
Anthropologists have as one of their special duties to science, the critical examination and broadening of theories that were originally developed in more culture-bound, usually Western contexts. Classical economic geography clearly improved when it was broadened to non-Western cases (e.g., Smith 1976). Anthropological study of ancient urban economics might broaden and enrich general economic theory by reexamining basic concepts, including money, price and value, credit and debt, savings and investment, and cycles of boom and bust.