Data and Empirical Methodology about tax--论文代写范文精选

当纳税人支付税收，在当地的银行,除了一张收据,两个额外的发票生成，其中一个是返回到税务部门,另一个是给财政部。后者的总数匹配两个部门的总量和实际银行转移到财政部。鉴于这些现有制度检查与实际收到钱,差异是相当罕见的。下面的essay代写范文进行详述。

Abstract 
The main source of the data for the project is quarterly administrative data on tax collections. Each quarter, as part of their normal reporting requirements, each circle inspector reports their revenue collected during the fiscal year cumulatively through the end of the quarter, which they compile from banks statements they have received. In addition, they also report their total assessed tax base before exemptions are granted (known as “gross demand”) and after exemptions have been granted (known as “net demand.”) These records are compiled separately for current year taxes and arrears. 

The excise and tax department uses these records for its internal monitoring purposes, but only has them in aggregated form (that is, each assistant supervisor creates a similar summary report for the circles he oversees; each supervisor creates a summary report for each assistant supervisor he oversees, and so on). We obtained and digitized the original quarterly reports at the circle level for each of the approximately 500 tax circles in Punjab for a total of 6 years (4 years prior to the project beginning and the 2 years the project was in place). These circle-level administrative records form the core of our analysis. One of the main advantages of performance pay in tax is that the outcome variable (tax revenues) is accurately observed, since it must correspond to money deposited in the bank and ultimately, received by the provincial Treasury department. When a taxpayer pays his tax due at the local bank, in addition to a receipt that he retains as proof of payment, two additional receipts are generated (the tax bill has three identical copies) that are collected by the bank. 

One of these is returned to the tax department and the other is given to the Treasury. The latter’s totals (at the district level) are then matched to both the department’s aggregates and also to the actual amount transferred by the bank to the Treasury. Given these existing institutional checks against actual money received, discrepancies are quite rare. However, to ensure that the department’s administrative data is correct at the circle-level as well (since the checks are usually run at a higher aggregation), we instituted an additional reverification program where we cross-checked the department’s administrative records against the bank records. 

This entailed selecting a subset of circles (done identically in treatment and control areas), obtaining the individual records of payment received from the bank for each property, and manually tallying the sums from the thousands of properties in each circle to ensure that it matched the department total. Each circle had about a 17% chance of being verified at some point. In general, we found virtually no systematic discrepancies between the administrative data we had received from the department and what we found in these independent verifications; the average difference between our independent verifications and what the circle had reported revealed under-reporting of -0.28%, or about zero. One complication is that circle boundaries are modified over time, as circles are merged and split to better reflect realities on the ground. 

In our data, out of the 482 circles present at the time of randomization, a total of 106 were affected by merges or splits throughout the 6 year period covered by our administrative data. To maintain consistency, we reconstruct the data at the level of the 482 circles present at the time of randomization. For those circles that merged prior to randomization, or split post randomization, one can simply add the two split circles together to obtain correct values for circles with randomization-era borders. For circles that split prior to randomization, or merged after randomization, we use the ratio of current year tax base net of exemptions (called “net demand” by the department) among the new and old circles reported in the quarters immediately before and after the split/merge to apportion the new circles to the randomization-era circles. The main results are qualitatively similar if we instead simply restrict analysis to the 376 circles that were unaffected, or aggregate all circle splits up. The administrative data also contains information on the identity of the inspector, which allow us to track if inspectors are relocated. We supplemented this by conducting a survey, each quarter, of the locations of all inspectors, constables, and clerks in the tax department.

Survey Data 
The second major data source is an independent property survey we conducted. This survey had three main purposes. First, it allowed us to obtain data on people’s interactions with the tax department, both in terms of their overall perceptions of the quality of this interaction and on corruption. Second, we obtained an independent assessment of the property’s characteristics (e.g. land area, covered area, location), which we could use to construct an independent assessment of the property’s valuation and compare to the department’s official assessments. Third, we could obtain information about the owners and property types which allow us to understand whether any observed impact of the schemes varied by the types of properties and owners. To do so, we surveyed approximately 16,000 properties. 

Properties were sampled in one of two ways. First, to obtain a random sample of all properties (including those not necessarily on the tax rolls), we created GIS maps of the circle boundaries for all 482 circles, and used GIS software to randomly select 5 points within each circle. Survey enumerators were given the latitude and longitude coordinates of each point, and used a handheld GPS device to locate that point on the ground. They then surveyed the property nearest that point, and selected 7 more properties nearby (chosen by walking left from the point and choosing every other property) of which an additional four were surveyed based on a randomization table . 

Once this was completed, we matched these properties to the property-level administrative data to obtain the corresponding administrative records for these properties. These properties, which we refer to as our “random sample,” represent a random sample of of 25 properties per circle. Second, since we were particularly interested in properties whose tax valuation had changed as a result of our treatment, we sampled properties directly off the separate tax lists that are maintained for newly assessed or re-assessed properties. Specifically, we randomly selected 10 properties in each circle among those that had been re-assessed during the FY11-12 and FY12-13 fiscal years, and then located these properties in the field and surveyed them. We denote this the “re-assessed” sample.15 The survey contained three main sets of questions. First, to compute an independent assessment of the property value, the surveyor noted all aspects of the property that feed into the property valuation, such as the land area, size of buildings, property use, etc.16 Second, to assess corruption, we ask a series of questions about unofficial payments. 

Given that respondents are often not comfortable revealing their own bribe payments, the key question we ask is the typical bribe that would need to be paid for a property similar to yours.17 Third, we collected a variety of characteristics about the owner, which we use to assess how the treatment differentially affects different types of households.18 The survey was conducted in two rounds, beginning at the end of the experiment. The first round was conducted primarily during June and July 2013 (with a few properties finished in August and September), and covered approximately half the circles (randomly selected). After a pause for the Eid holidays, the second phase of the survey was conducted from October 2013 to January 2014, and covered the remaining half of the circles. For subjective measures (e.g. bribes, customer satisfaction), we focus on the results from the first wave of the survey, when respondents would surely be answering for the correct time period when the treatments were in effect. For objective measures (e.g. accuracy of assessment, property characteristics), we use both survey waves.（essay代写)

51Due网站原创范文除特殊说明外一切图文著作权归51Due所有；未经51Due官方授权谢绝任何用途转载或刊发于媒体。如发生侵犯著作权现象，51Due保留一切法律追诉权。
更多essay代写范文欢迎访问我们主页 www.51due.com 当然有essay代写需求可以和我们24小时在线客服 QQ:800020041 联系交流。-X（essay代写)