Read an Excerpt
The Matrix of Policy in the Philippines
By Harvey A. Averch, John E. Koehler, Frank H. Denton
PRINCETON UNIVERSITY PRESSCopyright © 1971 The Rand Corporation
All rights reserved.
This book began with a misperception. Our research was originally intended to illuminate some of the links among the serious problems supposedly besetting the Philippines: crime, insurgency, political instability, and poor economic performance. However, when we began to examine the alleged problems and to search for the links, it became apparent that the problems themselves were exaggerated, or imaginary, or different in important ways from the manner in which they had been generally understood.
The misperception with which we began was widely shared. The view commonly upheld in the U.S. and Philippine press, the U.S. Congress, and even in scholarly work was predominantly negative. Even within the State Department, AID, and the Philippine government it was difficult to find a dissenting opinion. Individuals did disagree on particular points, and there was some difference between snap judgments and more thoughtful assessments, but it is fair to say that in all of these places the collective opinion held that the performance of the Philippines had been disappointing and that the future looked unpromising at best.
Forced to redefine the social and economic problems of the Philippines, we were led to a reconsideration of the ways in which policymakers — particularly U.S. policymakers — go about understanding events in other countries. Our book is not a proper study of the general process of perception in foreign affairs or of U.S. government intelligence and information. However, we have analyzed the deficiencies of the information with which Philippine policymakers must work, and we know that these deficiencies remain as the information is transmitted through U.S. reporting systems.
Whether the pathology of organizational intelligence that applies to the Philippines applies elsewhere is an open question. Studies that address the actual content and accuracy of information gathered by State, AID, the CIA, or the press are apparently scarce. So any generalization of our observations must rest on a line of reasoning more common to anthropology than to any of the other social sciences: this is the way it was in one exotic locale; we know of no reason why this place should be unique; it would be worth while to investigate whether or not the problems are general.
The Kind of Information Produced
Every year U.S. agencies in the Philippines produce an enormous quantity of reports and analysis. From what we saw, much of this reporting and analysis is very good. However, none of the analysis employs any quantitative tools, and much of the information is taken unexamined from Philippine sources. These two characteristics of the typical reporting style account for the failure of intelligence we have been discussing.
That the foreign service does not use any of the quantitative tools of modern behavioral science is a commonplace observation. In fact, the tradition and culture of the foreign service have been characterized as positively hostile to such methods, favoring the ideal of the expert so deeply immersed in a situation as to become a part of it and able to analyze problems almost intuitively.
Much useful reporting and analysis can be performed without reference to any of the tools of statistical inference and without any rigorous definition and testing of formal models, verbal or mathematical. But failure to apply quantitative methods at some point carries several hazards. One arises from the difficulty of aggregating intuitive perceptions. For example, an observer who saw numerous armed guards in Manila, read about insurgent incidents in Pampanga province, and talked with pessimistic local politicians would conclude that the country was in bad shape. Another observer who spent his time in a peaceful provincial city and talked with local Jaycee boosters would conclude the contrary. Without some quantitative test, there is no way to decide which observer saw the "true" Philippines. Furthermore, statistical techniques are mandatory for any analysis that hopes to capture the diversity of the Philippines. How else could we find answers to questions such as "Are parts of the country being left far behind the rapidly developing areas?" and "Are the dissidents found in Pampanga likely to enjoy success elsewhere ?"
Ultimately, nonquantitative interpretation of events in a country must rest on nonreproducible intuition. This may be highly accurate; a skilled and perceptive observer will be able to answer some questions quite well. The problem is to distinguish the skilled and perceptive from the blind and ignorant. If at some point arguments rest on a statement such as "I can't explain exactly why, but I'm sure that's the way it is," then an outsider is left powerless to check the statement without spending years in becoming immersed himself. In short, intuitive analysis carries with it no mechanism for the self-correction of error.
Self-correction is made even more unlikely by lack of independent sources of data. Much of the information reported through U.S. channels originates in the reporting agencies of the Philippine government. This is, of course, most characteristic of economic and social information; U.S. agencies have no capability to generate an independent estimate of GNP or unemployment or school enrollments without reference to official Philippine statistics. Political reporting may contain a large dose of independent observation, but here again at least a part of what is reported is likely to be the view of prominent Filipinos rather than independently gathered information.
Reliance on host-country sources for information implies that the views of government officials and the Manila elite will be reported quite accurately. For many purposes, that is precisely what policymakers want to know. But the corollary of this proposition is that if the government and the elite are mistaken in their judgments of conditions in the rest of the country, their mistakes will be shared.
How to Improve Reporting Systems
Our image of how reporting systems could be improved is inevitably colored by the experience of producing this book. To keep the record straight, we must confess that the method described here and elaborated in subsequent chapters just grew; it was not carefully plotted in advance. In retrospect, it appears to have been a good way to proceed. The style of quantitative investigation we promote here is not in any sense a substitute for the detailed reporting at the operational level that is now conducted. It does, however, offer ways to construct context more systematically than has been done before and to check periodically on the image of a country.
The prescription we offer has three directives: be quantitative, get inside the information systems, use multiple tests and new measures. The commitment to quantitative analysis is the fundamental feature. This commitment carries with it a focus on behavior that is relatively well measured or measurable. Thus in our discussions of elections we focus on voting behavior rather than on the alleged conspiracies of the Filipino elite. (The elite, as we have already noted, is quite well looked after by the present reporting systems.) The use of quantitative tools introduces a self-correction mechanism, because it is easier for a critic to detect error in the analysis. Models have to be made explicit, and test procedures are widely known. Anyone with the same data can check to see if the results are true or whether an alternative formulation of a model is superior.
Since information typically comes from the host country's reporting systems, it is important to get inside the system, to understand precisely where data come from, what they mean, and how they are gathered. In the Philippines, different government agencies often report contradictory data about what are ostensibly the same phenomena. Sometimes there are real errors, as we shall see in the data on crime. More generally, however, different agencies are really reporting on different things but giving them the same name. For example, alternative estimates of employment in manufacturing are produced from the economic census and the population census. The latter is three times as large as the former. The reason for the discrepancy becomes clear when we understand how the data are being gathered. The day-to-day reporting procedures of U.S. government agencies really assume that data that bear the same name in two countries measure the same characteristic, so that a sensible comparison can be made between, for example, values of some variable in Thailand and in the Philippines. Obviously, without the capability to analyze rigorously just how the numbers are being generated in both countries, the comparison is likely to be misleading.
Finally, given the uncertainties inherent in analysis, it is important to try to use different data and methods — preferably totally independent — to test for the existence of the same phenomena. In our analysis of politics, for example, we use reported vote and various province-level data to uncover some of the factors important in Philippine elections. This approach suffers from defects related to problems of aggregation and the types of data that are available at this level. The basic hypotheses are then checked by survey data. Survey data suffer inherent problems of response bias, control of field work, and reliability of coding. Their problems, however, are different from those of the aggregate data. If both sets of data support the same general image of the world, we can have higher confidence in it."
The Plan of the Book
The line of argument is continuous from Chapter 3 to Chapter 7. We are concerned with the prospects for political modernization, economic progress, and social stability in the Philippines. Chapter 3 considers the opinions of the public, politicians, and bureaucrats toward the performance of the political system and toward each other. All three groups have realistic expectations of each other, and their needs interlock so closely that the odds are against any sharp change.
Having established the context of attitudes within which political action occurs, we proceed to an analysis of Philippine elections in Chapter 4. Election outcomes, we found, are determined more by ethnic solidarity, local political control, and the delivery of pork barrel than by any issues arising out of poverty, unemployment, or tenancy. Our search for a group that is responding to programmatic appeals goes unrewarded, with the single exception of the rapidly growing but only slightly atypical modern labor force.
In Chapter 5 we look at the relations between politics and the economy. Despite the political motivation of much of government economic policy, the economy has turned in a respectable performance over the last two decades, better than usually thought. The demands of politics, however, have built a pattern of lurching, uneven growth into the economy. And although it has done reasonably well, political constraints may keep it from ever achieving a spectacular rate of growth.
Chapter 6 considers the problem of crime. The perception of "mounting crime" is an error, based on changes in reporting procedures and simple arithmetic mistakes. Nonetheless, the level of violence in Philippine society is generally high, yet it varies markedly across the country. Some violence is related to politics; some appears to be determined by the same ethnic factors we found to be important in the analysis of voting; some appears to come from urbanization. There is no evidence that Philippine crime should be considered to flow from general social deterioration.
Yet, if the society appears to be as stable as this analysis suggests, how do we explain the persistence of dissidence in Central Luzon ? Chapter 7 looks for the determinants of dissident success and tries to weigh the relative importance of social protest and simple terror. Here as in other fields, the principal obstacles to government success appear to be the requirements of politics.
Before we begin the analysis, we need a better understanding of the structure of the country and its recent historical evolution. Chapter 2 describes the overall structure of the Philippine social system, the context in which perceptions and attitudes are formed. The country is a patchwork of ethnic, political, religious, and economic groups. What then do we mean when we talk about the Philippine nation? It is customary to begin the analysis of a country's social and economic structure with a historical and geographical overview. We shall not follow this procedure, since the basic descriptive work is available elsewhere. Instead, we shall attempt to summarize quantitatively the pattern of constants and variables over the last thirty years.CHAPTER 2
Philippine Geography and Society: Factor Analysis
The Philippines stretches more than 800 miles from Aparri in the north on the flood plain of the Cagayan river to JoIo in the south. Contained within the country is a range of geography, climate, culture, and economic structure as broad as we can find anywhere. Christianity and Islam are the two major religions; there are 11 major languages and over 70 minor languages; and styles of living range from the Ifugao of Central Luzon who occasionally revive the headhunting customs of their ancestors to Manilans as at home in Paris and Washington as in Makati. Figure 1 shows the locations of the major language groups.
Since the country is diverse in so many dimensions, we use factor analysis — a data-description technique — to summarize the relations among social, political, economic, and cultural characteristics in different parts of the country. We know that each area has a few fundamental characteristics (ethnic composition, economic structure, and so on). The large number of attributes for which we can obtain measures are all highly correlated with these few fundamental characteristics. For example, if we know that urbanization, high incomes, and the proportion of the labor force in manufacturing all go together, then we can speak of a composite of these indicators, which we may want to call "development." Our interest is to provide a compact portrait of the country using these composite indicators as a context for later analysis.
The substantive idea behind factor analysis is that different phenomena may be part of the same underlying cluster or pattern. For example, by economic development we mean not just increases in GNP per capita, but a set of political and social changes we believe are related: urbanization, education, communication, emergence of a modern labor force — the list could be extended indefinitely. Without some formal device for reducing the number of variables at hand, we cannot readily describe or grasp the nature of complicated processes. Factor analysis is one such formal device. It is a systematic way of reducing the dimensions of a problem and of revealing clusters of associated variables.
The technical vocabulary of factor analysis is unfamiliar but not difficult. With a few definitions the results of a factor analysis can be read quickly by laymen and policymakers. Results are usually displayed in table form with such labels as factors, loadings, and communalities. A factor means an underlying pattern or association among a set of variables. A factor loading means essentially the correlation between a variable and a factor. For example, urbanization might load (correlate) +.9 on a "development" factor, suggesting that urbanization and development are associated. A communality means the sum of the squares of the factor loadings. This sum shows the total contribution of the factors operating together in explaining the (standardized) variance of a given variable. It is analogous to the multiple correlation coefficient in regression analysis.
A factor score is the ranking or rating of cases or observations on all the variables entering a pattern or factor. Factor scores have a standard normal distribution, so 80 percent of the scores will lie in the interval [— 1.28, 1.28]. If high GNP per capita, urbanization, communication, and education were part of a general development factor, then the United States would score high with, say, a score of 2, Japan might be in the middle with a score of 0, and Nepal might be near the bottom with a score of — 2. The factor scores permit us to make rankings of geographic units such as provinces, states, or nations, using all the available information in a systematic way. Scores can be plotted on maps rather than presented as lists.
Because we wished to analyze historical change in the geographic distribution of economic and political activity and of dissidence, we selected provincial data for three different census years — 1938, 1948, and 1960. Sometimes important variables were measured only for other years. We added some post-1960 data, such as 1965 voting patterns, to see their relation to the evolving economic and political structure. The three years taken together give us some idea of change and constancy in the Philippine social system.
Excerpted from The Matrix of Policy in the Philippines by Harvey A. Averch, John E. Koehler, Frank H. Denton. Copyright © 1971 The Rand Corporation. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.